f



RE: NTP ( Network Time protocol ) With Ingres #2

> At 11:34 AM 7/18/2003 +0200, Bruno Wipier wrote:
> >Does somebody use NTP ( ntpdate or xntpd ) with Ingres. 
> Officialy NTP is not
> >compatible with Ingres.
> >
> >Thanks.
> >
> >Bruno.
> 
> Bruno,
> 
> I guess I'm a bit flustered.  Is there some list of
> UNIX utilities that are or are not "compatible" with
> Ingres?
> 
> NTP merely keeps your clock in sync, usually within
> milliseconds of the 'master'.
> 
> Is Ingres not compatible with the correct time?  ;-)

We don't support changing the system clock whilst ingres is running, especially if it goes backwards. It will affect timestamps on journal records as well as any user data created with date('now') etc. 

From the man page of ntpdate, if the time is out of sync by < 0.5 sec then it will adjust it by calling adjtime, which doesn't actually change the clock it merely speeds it up or slows it down slightly until the time is back in sync. This method should be ok because it means that the time is always increasing as it should.

If the difference is > 0.5 sec then it will set the time by calling settimeofday. This should be avoided while ingres is running, and if the time is set backwards then ingres should really be kept down long enough for the new clock time to overtake the old. 

The point about 6.4 and DST is a good one. 

HTH
Paul Mason

0
Paul
7/18/2003 4:32:35 PM
comp.databases.ingres 5857 articles. 0 followers. specially (35) is leader. Post Follow

4 Replies
1992 Views

Similar Articles

[PageSpeed] 20

At 04:37 PM 7/18/2003 +0100, Mason, Paul wrote:
> > At 11:34 AM 7/18/2003 +0200, Bruno Wipier wrote:
> > >Does somebody use NTP ( ntpdate or xntpd ) with Ingres.
> > Officialy NTP is not
> > >compatible with Ingres.
> > >
> > >Thanks.
> > >
> > >Bruno.
> >
> > Bruno,
> >
> > I guess I'm a bit flustered.  Is there some list of
> > UNIX utilities that are or are not "compatible" with
> > Ingres?
> >
> > NTP merely keeps your clock in sync, usually within
> > milliseconds of the 'master'.
> >
> > Is Ingres not compatible with the correct time?  ;-)
>
>We don't support changing the system clock whilst ingres is running, 
>especially if it goes backwards. It will affect timestamps on journal 
>records as well as any user data created with date('now') etc.
>
> From the man page of ntpdate, if the time is out of sync by < 0.5 sec 
> then it will adjust it by calling adjtime, which doesn't actually change 
> the clock it merely speeds it up or slows it down slightly until the time 
> is back in sync. This method should be ok because it means that the time 
> is always increasing as it should.
>
>If the difference is > 0.5 sec then it will set the time by calling 
>settimeofday. This should be avoided while ingres is running, and if the 
>time is set backwards then ingres should really be kept down long enough 
>for the new clock time to overtake the old.
>
>The point about 6.4 and DST is a good one.
>
>HTH
>Paul Mason

*Sigh*

This is a difficult restriction.   Many of us need our systems to be
kept in sync, and almost every platform I have used drifts seconds per
day.

It seems to me (maybe too simplisticly) that this should work:

   - All internal dates within Ingres should use GMT.

   - All "sequenced" things within Ingres should not rely on time but instead
     use a generated sequenced key.

The UNIX kernel guys have dealt with and solved time issues decades ago.
All their solutions are in the public view.

It just isn't realistic to let your clock drift higglety-pigglety.

Just my opinion ...







Michael Leo            mleo@cariboulake.com        Java, J2EE, .NET
Caribou Lake Software  http://www.cariboulake.com  Oracle, Ingres,

So much style without substance, so much stuff without style.
It's hard to recognize the real thing. It comes along once in a while.
                              - Grand Designs, Power Windows, Rush

0
mleo
7/18/2003 5:58:55 PM
Thanks for this informations but i don't know what to do ! I think i'm going
to run "ntpdate" when the database has little activity.( the night after a
ckpdb ).

Bruno.



"Paul Mason" <paul.mason19@NOSPAMvirgin.net> a �crit dans le message news:
v8tghvsc32tbggrliam9t71cn0g598qmp2@4ax.com...
> On 18 Jul 2003 12:58:55 -0500, mleo@cariboulake.com (Michael Leo)
> wrote:
>
> >At 04:37 PM 7/18/2003 +0100, Mason, Paul wrote:
> >> > At 11:34 AM 7/18/2003 +0200, Bruno Wipier wrote:
> >> > >Does somebody use NTP ( ntpdate or xntpd ) with Ingres.
> >> > Officialy NTP is not
> >> > >compatible with Ingres.
> >> > >
> >> > >Thanks.
> >> > >
> >> > >Bruno.
> >> >
> >> > Bruno,
> >> >
> >> > I guess I'm a bit flustered.  Is there some list of
> >> > UNIX utilities that are or are not "compatible" with
> >> > Ingres?
> >> >
> >> > NTP merely keeps your clock in sync, usually within
> >> > milliseconds of the 'master'.
> >> >
> >> > Is Ingres not compatible with the correct time?  ;-)
> >>
> >>We don't support changing the system clock whilst ingres is running,
> >>especially if it goes backwards. It will affect timestamps on journal
> >>records as well as any user data created with date('now') etc.
> >>
> >> From the man page of ntpdate, if the time is out of sync by < 0.5 sec
> >> then it will adjust it by calling adjtime, which doesn't actually
change
> >> the clock it merely speeds it up or slows it down slightly until the
time
> >> is back in sync. This method should be ok because it means that the
time
> >> is always increasing as it should.
> >>
> >>If the difference is > 0.5 sec then it will set the time by calling
> >>settimeofday. This should be avoided while ingres is running, and if the
> >>time is set backwards then ingres should really be kept down long enough
> >>for the new clock time to overtake the old.
> >>
> >>The point about 6.4 and DST is a good one.
> >>
> >>HTH
> >>Paul Mason
> >
> >*Sigh*
> >
> >This is a difficult restriction.   Many of us need our systems to be
> >kept in sync, and almost every platform I have used drifts seconds per
> >day.
>
> Why is that btw? Why is it that a cheap digital watch can keep time
> better than an expensive computer? I know it's a cliched joke but does
> anyone know why?
>
> >
> >It seems to me (maybe too simplisticly) that this should work:
> >
> >   - All internal dates within Ingres should use GMT.
>
> They do.
>
> >
> >   - All "sequenced" things within Ingres should not rely on time but
instead
> >     use a generated sequenced key.
>
> They do. The logging system uses Log Sequence Numbers which do not
> depend on the system clock.
>
> It's just that journal records also have a timestamp so you can run
> auditdb and rollforward to a point in time. If, because of time
> adjustments, we can have multiple journal records for the same point
> in time then it makes that tricky.
>
> Maybe we should have rollforward based on LSNs but that could cause
> confusion - what do you think?
>
> >
> >The UNIX kernel guys have dealt with and solved time issues decades ago.
> >All their solutions are in the public view.
> >
>
> I don't think they apply in the same way. Having said which I'm not a
> kernel expert maybe there's something which is analogous to
> rollforward.
>
> >It just isn't realistic to let your clock drift higglety-pigglety.
>
> Well then you have two choices - shut down ingres before re-syncing it
> or do it with ingres running and risk problems if you need to
> rollforward.
>
> I should point out something about what "supported" means. It mostly
> means "is not guaranteed to work 100% of the time" - it doesn't mean
> Tech Support won't take your call. We'll *always* try to help whatever
> the situation.
>
>
> --
> Paul Mason


0
Bruno
7/21/2003 8:57:58 AM
Hi Bruno,

>Thanks for this informations but i don't know what to do ! I think i'm going
>to run "ntpdate" when the database has little activity.( the night after a
>ckpdb ).

As explained earlier in the thread, the main risk of NTP is stepping
the clock backwards. I think there is no way to 100% guarantee that
this will never happen. You can influence it, however.

Your strategy is to take the risk of stepping backwards at a moment
when the system is quiet, thus limiting the effect of the clock being
set back. But, because you do this only once a day, the system clock
had a full day to drift, and the chance of a large difference between
system time and network time has increased by that decision.

If you run a NTP server on your Ingres server, the NTP server will
build a history of the drift (man ntp.drift). Since most often a
system's drift is linear (specially when the computer is in a
conditioned environment with a constant temp), the NTP server can keep
the time very accurate even if it cannot reach a higher stratum NTP
server for a few days (it predicts and corrects the drift based on
history records).

So, when the drift is known (if the NTP server has run for a few
days), the possibility of a large difference between system time and
network time is much smaller. Therefore, I should prefer a correctly
configured NTP server over ntpdate.

Correctly configured? Yes that is important, if an NTP server has not
enough reliable sources (higher stratum NTP server) it can also step
the clock, something we want to avoid at all times. My advice would be
to configure at least three other NTP sources.

You can even decrease the chance of stepping by:
- using NTP servers reachable over different networks
- using also a DCF receiver or a GPS receiver as a source, some GPS
based clocks support NTP out-of-the-box

So, depending on your needs (how important is the accurate time for
your app, how much do you want to spend) you can make NTP very safe by
avoiding single points of failure. But, if your local NTP server has
lost all of its time sources for a long time, at the moment it can
reach one of these sources again, the clock might step. But you can
monitor the loss of all connections, and decide to bring down your
local NTP server when this ever happens. Bringing it up again must be
done when no servers are running (e.g. single user mode), because it
is likely to step the clock.

Never rely on a single NTP source, even not on a very accurate GPS
receiver. If it fails for a relevant period of time and comes back
after that, the chances of stepping are realistic. Given enough
sources, a NTP server is able to decide if clock sources are in error,
and the chances of stepping are very small. If they are small enough
for your business is up to you to decide.

Hope this helps,
Eric
0
E
7/22/2003 7:00:04 PM
Many thanks for this advice.

Bruno.
-
"Eric Schreuder" <E.Schreuder@imn.nl> a �crit dans le message news:
3f1d8207.3884755@news.nl.net...
> Hi Bruno,
>
> >Thanks for this informations but i don't know what to do ! I think i'm
going
> >to run "ntpdate" when the database has little activity.( the night after
a
> >ckpdb ).
>
> As explained earlier in the thread, the main risk of NTP is stepping
> the clock backwards. I think there is no way to 100% guarantee that
> this will never happen. You can influence it, however.
>
> Your strategy is to take the risk of stepping backwards at a moment
> when the system is quiet, thus limiting the effect of the clock being
> set back. But, because you do this only once a day, the system clock
> had a full day to drift, and the chance of a large difference between
> system time and network time has increased by that decision.
>
> If you run a NTP server on your Ingres server, the NTP server will
> build a history of the drift (man ntp.drift). Since most often a
> system's drift is linear (specially when the computer is in a
> conditioned environment with a constant temp), the NTP server can keep
> the time very accurate even if it cannot reach a higher stratum NTP
> server for a few days (it predicts and corrects the drift based on
> history records).
>
> So, when the drift is known (if the NTP server has run for a few
> days), the possibility of a large difference between system time and
> network time is much smaller. Therefore, I should prefer a correctly
> configured NTP server over ntpdate.
>
> Correctly configured? Yes that is important, if an NTP server has not
> enough reliable sources (higher stratum NTP server) it can also step
> the clock, something we want to avoid at all times. My advice would be
> to configure at least three other NTP sources.
>
> You can even decrease the chance of stepping by:
> - using NTP servers reachable over different networks
> - using also a DCF receiver or a GPS receiver as a source, some GPS
> based clocks support NTP out-of-the-box
>
> So, depending on your needs (how important is the accurate time for
> your app, how much do you want to spend) you can make NTP very safe by
> avoiding single points of failure. But, if your local NTP server has
> lost all of its time sources for a long time, at the moment it can
> reach one of these sources again, the clock might step. But you can
> monitor the loss of all connections, and decide to bring down your
> local NTP server when this ever happens. Bringing it up again must be
> done when no servers are running (e.g. single user mode), because it
> is likely to step the clock.
>
> Never rely on a single NTP source, even not on a very accurate GPS
> receiver. If it fails for a relevant period of time and comes back
> after that, the chances of stepping are realistic. Given enough
> sources, a NTP server is able to decide if clock sources are in error,
> and the chances of stepping are very small. If they are small enough
> for your business is up to you to decide.
>
> Hope this helps,
> Eric


0
Bruno
7/23/2003 11:45:58 AM
Reply:

Similar Artilces:

RE: NTP ( Network Time protocol ) With Ingres
Bruno, Our servers (HPUX10.20 OpenIngres 1.2) have been running NTP for years without a problem. (Currently version ntp-4.0.99j) Maybe we have been lucky. Richard > ---------- > From: Bruno Wipier[SMTP:bruno.wipier_nospam_@free.fr] > Sent: 18 July 2003 10:34 > To: info-ingres@ams.org > Subject: NTP ( Network Time protocol ) With Ingres > > Does somebody use NTP ( ntpdate or xntpd ) with Ingres. Officialy NTP is > not > compatible with Ingres. > > Thanks. > > Bruno. > > > > ________________________________________________________________________ > This e-mail has been scanned for all viruses by Star Internet. The > service is powered by MessageLabs. For more information on a proactive > anti-virus service working around the clock, around the globe, visit: > http://www.star.net.uk > ________________________________________________________________________ > ________________________________________________________________________ This e-mail has been scanned for all viruses by Star Internet. The service is powered by MessageLabs. For more information on a proactive anti-virus service working around the clock, around the globe, visit: http://www.star.net.uk ________________________________________________________________________ ...

Re: NTP ( Network Time protocol ) With Ingres #4
Paul Mason wrote: > Why is it that a cheap digital watch can keep time better than > an expensive computer? Well, a cheap digital watch does not have so much to think about! (You've just reminded me of Mr. Prosser, for some reason ( <http://tinyurl.com/hkde> ) ). Still, the basic principle is the same - oscillator frequency variation will make some difference. There may be some issue with lost clock interrupts, AFAIR. Each clock interrupt causes the kernel to increment the time (by 10 milli- seconds, 1/100 second?), so a higher priority interrupt causing a missed clock interrupt would result in time slippage. This is all a long time ago, and I can't remember whether that is for an old mainframe, or a Sun system. It probably doesn't matter, anyway - my 1st. Law of Computing - "all computers are the same computer". Oh, I think temperature has some issue on clock speed, too. This *is* irritating, I may have to go and look it up! -- Erratic News-Swervers R us! Ingres, yeah baby, yeah!!! Oh, behave!! ________________________________________________________________ Sent via the PAXemail system at paxemail.com ...

RE: NTP ( Network Time protocol ) With Ingres #7
> From: Michael Leo > > > > > >This is a difficult restriction. Many of us need our > systems to be > > >kept in sync, and almost every platform I have used drifts > seconds per > > >day. > > > >Why is that btw? Why is it that a cheap digital watch can keep time > >better than an expensive computer? I know it's a cliched > joke but does > >anyone know why? > > I've always wondered this myself. I'm and Electrical Engineer > and used to design a bit of hardware myself. I don't get why this > happens. Drives me wacko. > It looks like it's a temperature thing, as suggested by Richard, http://www.futurlec.com/News/Dallas/Crystal.html <snip> > > > - All "sequenced" things within Ingres should not rely > on time but > > instead > > > use a generated sequenced key. > > > >They do. The logging system uses Log Sequence Numbers which do not > >depend on the system clock. > > Cool. Makes me feel better. > > >It's just that journal records also have a timestamp so you can run > >auditdb and rollforward to a point in time. If, because of time > >adjustments, we can have multiple journal records for the same point > >in time then it makes that tricky. > > > >Maybe we should have rollforward based on LSNs but that...

RE: NTP ( Network Time protocol ) With Ingres #8
This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_001_01C35026.97706250 Content-Type: text/plain The only thing wrong with daylight saving in the uk is the fact the don't jump 2 hours ahead! Who cares if it's light at 4.30 am, apart from the whinging farmers of course. Oh, and not forgetting that daylight saving is an absurd concept. It's not as though you can put it in the bank and withdraw it when you like "Yes bank clerk, I'd like to withdraw an extra hour of daylight in my garden tonight". Of course, in the UK it's referred to as British Summer Time, which makes you want to break into an MC Hammer song just without the baggy trousers and cucumber sandwiches instead... -----Original Message----- From: Karl & Betty Schendel [mailto:schendel@kbcomputer.com] Sent: 21 July 2003 17:27 To: info-ingres@ams.org Subject: Re: NTP ( Network Time protocol ) With Ingres At 11:07 AM -0500 7/21/03, Michael Leo wrote: > >Can it be that tough to solder a digital watch (joking) >on the corner of the motherboard and not let it be bothered >by interrupts. How much are digital watches nowadays? >How much are 8-way servers? Actually I think Sun's (at least) are supposed to have some kind of RTC clock chip that does just that. I dunno why they don't seem to work very well. > >Then the...

Re: NTP ( Network Time protocol ) With Ingres #6
By the way, even though most servers I meet can't keep time to save a life, my 4-year-old Powerbook G3 laptop will generally hold time to within a few seconds every couple months. Even the ancient (9-year-old) Performa 6110 that we just retired could keep time better than my Ultra 2 can. -- Karl R. Schendel, Jr. schendel@kbcomputer.com K/B Computer Associates www.kbcomputer.com Ingres, Unix, VMS Consulting and Training ...

RE: NTP ( Network Time protocol ) With Ingres #5
> From: Ingres, yeah baby!! [mailto:Use-Author-Address-Header@[127.1]] > Paul Mason wrote: > > Why is it that a cheap digital watch can keep time better than > > an expensive computer? > > Well, a cheap digital watch does not have so much to think > about! (You've just reminded me of Mr. Prosser, for some > reason ( <http://tinyurl.com/hkde> ) ). Still, the basic > principle is the same - oscillator frequency variation > will make some difference. There may be some issue with > lost clock interrupts, AFAIR. Each clock interrupt causes > the kernel to increment the time (by 10 milli- seconds, > 1/100 second?), so a higher priority interrupt causing a > missed clock interrupt would result in time slippage. > This is all a long time ago, and I can't remember whether > that is for an old mainframe, or a Sun system. It probably > doesn't matter, anyway - my 1st. Law of Computing - "all > computers are the same computer". Oh, I think temperature > has some issue on clock speed, too. This *is* irritating, > I may have to go and look it up! > Can't we just do something like hook up a cheap digital watch to motherboard so it has it's own internal 'time-server' that it can sync against periodically. Then all you'd need is the equivalent of ntpdate in firmware, set to run regularly enough so it never has to 'step' the time, give it a fr...

Re: NTP ( Network Time protocol ) With Ingres #3
Bruno Wipier wrote: > > Thanks for this informations but i don't know what to do ! I > think i'm going to run "ntpdate" when the database has little > activity.( the night after a ckpdb ). Hi Bruno, what is the scale of your clock drift? IMO "little and often" is the best policy for using ntpdate - you don't want to run it to correct large time differences, and, as it over-corrects anyway, it might be (very slightly) counter- productive to do so - for instance, if the time is 10 secs. slow, and you run ntpdate to correct this large inaccuracy, then you might end up with a time 5 seconds fast! Anyway, as I mentioned we run "ntpdate -Bs" daily on some quite busy systems, and I have never seen a problem caused by it. If I were you, I'd set it up to "step" the time at next boot, or when Ingres is stopped if this is otherwise acceptable, and then run -Bs every so often. It is well worth having at least two NTP servers as reference. -- Erratic News-Swervers R us! Ingres, yeah baby, yeah!!! Oh, behave!! ________________________________________________________________ Sent via the PAXemail system at paxemail.com ...

RE: [Info-ingres] Re: Ingres stand-by database #2
Well aware of the -b bugs. We have filed many issues and have them all resolved on our version. We don't xfer the ckp's because it is 60GB over a WAN. We also "zero-out" the .jnl files before and after the ones we want to rollforward. We do NOT monkey with the .cnf file, just transfer it. -----Original Message----- From: martin.bowes@ctsu.ox.ac.uk [mailto:martin.bowes@ctsu.ox.ac.uk] Sent: Thursday, June 09, 2005 4:13 AM To: Kale, Mark (STP) Cc: info-ingres@cariboulake.com Subject: Re: [Info-ingres] Re: Ingres stand-by database Hi Mark I think I may have missed a point here, but... You do a backup of your production database and then only shove the journals over to the recovery host - not the checkpoint you just took. Then, after monkeying with the config file, you use the journals to do a rollforwarddb -c -b? Why? Why don't you use the checkpoint you just made on the production host, transfer those checkpoint tar files to the DR host and do a normal recovery? What have I missed? BTW. The -b flag on rollforwarddb has been subject to some bugs. Specifically relating to whether or not it represents the transaction commit time - as it is supposed to. Marty > Chloe, > > Thought I would chime in. We have a drp box configured almost > identically to production. We have a cronjob that copies over journal > files (not yet copied) every 10 minutes to our drp server. We set a...

RE: [Info-ingres] Re: PHP times out waiting for an ingres connection #2
I tried altering the hosts file as you recommended but no change in response, so I'm not laughing. Still it takes 6 seconds for the connect. Trying ping from my NT client to my VMS server, responds immediately. Roy, can you reproduce the problem by you? On my NT I have: MS Windows server 2003 Enterprise edition II 2.6/0305 (int.w32/00) 10520 And my server is OpenVMS 7.3-1: II 2.0/0308 (axm.vms/00) Patch 10680 CA, as of yet, have not been able to produce a solution. Yehudah -----Original Message----- From: info-ingres-admin@cariboulake.com [mailto:info-ingres-admin@cariboulake.com] On Behalf Of Roy Hann Sent: Sunday 15 May 2005 13:50 To: info-ingres@cariboulake.com Subject: Re: [Info-ingres] Re: PHP times out waiting for an ingres connection "Yehudah Edelstein" <YEHUDAED@cellcom.co.il> wrote in message news:mailman.1115736482.25913.info-ingres@cariboulake.com... > Do you have timeout defined for the transaction? I have NT Ingres > client trying to connect to OpenVMS and it takes 6 seconds. The > slowness is on the NT Ingres. Just for a laugh, see if adding the name of the NT machine to its own drivers\etc\hosts file speeds up the time needed to make a connection. Roy Hann (rhann at rationalcommerce dot com) Rational Commerce Ltd. www.rationalcommerce.com "Ingres development, tuning, and training experts" _______________________________________________ Info-ingres mailing list In...

Re: [Info-Ingres] Upgrading from Ingres 2.6 to Ingres 2006- version9.2
I will be out of the office now until Wednesday 25th March, and will respond to your e-mail as soon as I can upon my return. If the matter is urgent, please raise a call via the ICT Helpdesk. This e-mail and any files transmitted with it are the property of the London Borough of Havering, are confidential, may be subject to legal privilege and are intended only for the person(s) or organisation(s) named above. Any unauthorised use, retention, distribution, copying or disclosure is strictly prohibited. If you receive this email in error, please notify the sender immediately and delete this e-mail from your system. WARNING: It is your responsibility to take all necessary steps to ensure this e-mail and any attachments to it are free from viruses. ...

[Info-ingres] RE: [ingres] [Info-ingres] Dates users were created in Ingres #2
Hi Troy, iidbdb journals are not that big. In one of my larger installations, 3 days worth of journals takes 112Mb for the transaction database and 200Kb for iidbdb. I think in Ingres 2.6 you can keep 100 checkpoints in the history. Paul -----Original Message----- From: Anderson, Troy E. [mailto:AndersonTE@bvsg.com] Sent: Monday, 17 January 2005 3:57 PM To: Paul Mason Cc: info-ingres@cariboulake.com Subject: RE: [ingres] [Info-ingres] Dates users were created in Ingres Hi Paul, Thanks for the information! I do have a question: I checkpoint the iidbdb daily with the -d flag, so the journal files don't exist for me that far back (to cover all of 2004). If I weren't using the -d flag, how far back would Ingres track Journal/Checkpoint information? Indefinitely? I have a limited amount of space in my jnl file system, so I doubt I could keep a year's work of jnl files for that database even if I weren't using the -d flag. Thanks, Troy -----Original Message----- From: Paul Mason [mailto:latepaul@gmail.com] Sent: Sunday, January 16, 2005 4:32 PM To: Anderson, Troy E. Cc: info-ingres@cariboulake.com Subject: Re: [ingres] [Info-ingres] Dates users were created in Ingres Anderson, Troy E. wrote: > Hi, > > Does anyone know if Ingres tracks the date that a user was created (in > accessdb)? Is that contained in a system catalog, for example? I'd > like to figure out all new user accounts cr...

RE: [Info-ingres] Ingres stand-by database #2
I am not doing a creeping one. It just allows a copy of the up-to-date journals and last ckp file etc. to be ready for A rollforward when required. -----Original Message----- From: martin.bowes@ctsu.ox.ac.uk [mailto:martin.bowes@ctsu.ox.ac.uk] Sent: Wednesday, 8 June 2005 8:12 p.m. To: Robert Allely; 'chloe.crowder@bl.uk' Cc: info-ingres@cariboulake.com Subject: Re: [Info-ingres] Ingres stand-by database Hi Robert, Chloe etc. I'm not sure if I have a full grasp on whether or not Robert was doing a 'creeping' rollforward or not, but just in case, I'd like to get on the Ol' Soap Box for a moment. Its one thing to set up a DR system using rollforwarddb from a checkpoint/journal etc. Thats a perfectly simple and natural thing to do. But Its quite another to use rollforwarddb from some point and then transfer over more journals and do a 'creeping' rollforwarddb -c +j '#f' - b... [-norollback] ....And then automate the process, ....allowing for errors such as: * the rollforwarddb being locked out by another process, * the possibility of a database being flagged as inconsistent on the DR site - which is very probable. * A breakdown in the transport of a journal or other essential file between the hosts. scp/rcp/ftp failure. * A journal being written to by the primary host archiver as it was transferred to the DR host. Whats that (unsupported) trace point to force ...

RE: [Info-ingres] Re: Ingres/ICE (3.0.2) and Apache 2.0.54 #2
The example PHP file is available via http://www.ingres.co.uk/files/php/blob_read.phps and not the url posted earlier. ...

NTP ( Network Time protocol ) With Ingres
Does somebody use NTP ( ntpdate or xntpd ) with Ingres. Officialy NTP is not compatible with Ingres. Thanks. Bruno. At 11:34 AM 7/18/2003 +0200, Bruno Wipier wrote: >Does somebody use NTP ( ntpdate or xntpd ) with Ingres. Officialy NTP is not >compatible with Ingres. > >Thanks. > >Bruno. Bruno, I guess I'm a bit flustered. Is there some list of UNIX utilities that are or are not "compatible" with Ingres? NTP merely keeps your clock in sync, usually within milliseconds of the 'master'. Is Ingres not compatible with the correct time? ;-) Or are you running Ingres 6.4 and you are manually dealing with Day Light Savings Time? In that case, you might be fudging with the system clock and/or time zone. I wouldn't run NTP in that case, unless you learn to configure it to your rules. By the way, DLST is evil. Where, pray tell, is this "official" non-compatibility posted? Cheers, Michael Leo mleo@cariboulake.com Java, J2EE, .NET Caribou Lake Software http://www.cariboulake.com Oracle, Ingres, So much style without substance, so much stuff without style. It's hard to recognize the real thing. It comes along once in a while. - Grand Designs, Power Windows, Rush ...

RE: [Info-ingres] Re: How Different is Ingres? #2
Hi there, >I can't see why a similar locking option couldn't be added to Ingres in >time?? > Hmmm.... way back when, Postgres had a feature called "time travel", where you could consciously step back to older versions of rows of data to see how it used to look. At this time, it still used conventional locks. Around version 6.5 (6.5.2 ?) the time travel feature was dropped and converted into MVCC, and traditional spinlocks were dropped as the default mechanism (you can still get them by doing "select for update", IIRC) The thing is that Postgres gets MVCC by never updating a row in place; it creates a new copy of the row with a new version number to supplant the old row for new transactions. This means that after bulk updates, say, your table expands by the number of rows you updated. Copious use of the "vacuum" command (or "vacuum analyze" for optimizedb-like stats gathering) keeps things in check. So it might not be the easiest thing to add architecturally - not a simple bolt-on anyway, I'd reckon. It *might* need fiddling with the internal representation of rows in tables to get it working properly, but I haven't looked at the Postgres source in ages and never at the Ingres source in anger ! Cheers, - Tony (not as mad as Roy makes out - not all the time, anyway...) __________________________________________________________________ Switch to Netscape Internet Service. As low as $9.95 ...

Re: [Info-ingres] Re: What animal should Ingres be? #2
Hi there, Emiliano <Emiliano@Iris-Advies.nl> wrote: >I don't really care whether it's java, or python, or C#, or Tcl or even >lisp; it's just that many of the developers around me that have grown up > on the 'regular' procedural and/or OO languages find Ingres' SP >language arcane. > I don't really care what language folks want to use either; that's why I think the choice should be left open. Individual sites will no doubt choose a standard, but that standard may vary. I've never found the syntax of the procedure language that arcane; an if statement is pretty much an if statement wherever you go really (unless it's occam, I suppose), although the row producing stuff has, like session global temporary tables, a syntax surely designed to induce RSI. Assuming the path of least resistance, that is, beefing up what we've already got rather than total reinvention, I'd like ... - A debugger. Failing that, just a decent trace mechanism. Single step would be lovely. I know performance would be crap, but this is for development here. - How's about being able to retrieve from Ingres the procedure you put in - complete with the whitespace you put in to make it readable, and the comments to make it clear what you intended (even, or especially, when that's not what you actually *did* ;) - Some kind of version control would be wonderful. - I'd like the restrictions that are...

Re: [Info-ingres] Re: What animal should Ingres be? #2
Hi there, "Paul Andrews" <ac297@dial.pipex.commmmm> wrote: >Oops, the database procedural language debate has turned into some kind of >academic discussion and is in danger of dissappearing up it's own posterior. > Isn't that pretty much what these types of discussions are for ? >Here's some ill-informed thoughts of my own. > >We're debating something, not even sanctioned by ingres corp. Welcome to open source. With the source available, we don't really need to ask for permission about this sort of stuff. The open source world works by having (near) endless discussions like this, then usually 2 or 3 camps will form, each implementing their view of nirvana. Then the community chooses the one that works best for them. This applies to everything, from operating system virtual memory management up. Your chances of support from Ingres for your experimental DBMS server are pretty minimal (read: 0) though ... >Even if it were to go anywhere I suspect the development budget would be >minimal and I >also suspect it's too big a job for enthusiastic ingres open-source >developers (if there are any). The problem for us is that Ingres is a complex beast, nowhere near as well publicly understood as (say) the Linux kernel, so there probably aren't enough skilled and motivated souls to set up those differing camps. >If it were that easy CA might have gone forward in this direction....

Re: [Info-ingres] Re: What animal should Ingres be? #2
Hi Mike, Michael Leo <mleo@cariboulake.com> wrote: [ snippage ] > >Give me the gun. I will decide how to use the safety. > >I've seen enough people shoot themselves in the foot using >Ingres' neutered stored procedure language. On the other hand, if they're shooting themselves in the foot with a pop-gun, maybe handing them a pump-action shotgun isn't a great idea ;) > Are you proposing >an even LESS useful language to somehow prevent this? > Well, I'm still preferring the idea of letting the sites decide by implementing an open framework. And in passing, I notice that there are two implementations of Java as a procedure language in PostgreSQL (PL/Java and pl-j), so we could all be happy. I get R, y'all get Java, everybody's happy ... >Mikey > Cheers, - Tony __________________________________________________________________ Switch to Netscape Internet Service. As low as $9.95 a month -- Sign up today at http://isp.netscape.com/register Netscape. Just the Net You Need. New! Netscape Toolbar for Internet Explorer Search from anywhere on the Web and block those annoying pop-ups. Download now at http://channels.netscape.com/ns/search/install.jsp ...

RE: [Info-ingres] Re: Ingres article #2
At 9:59 AM +1100 2/3/2006, Paul White wrote: >When CA announced they were moving their internal systems to SAP the >obvious question was asked - are they going to run SAP over ingres? I >wonder if this is one of the drivers. >Paul > This is strictly hearsay, but I seem to recall hearing that CA decided to run SAP on DB/2. They have some hard time constraints with very VERY severe penalties for missing deadlines. I have no idea how seriously SAP on Ingres was discussed at the time. Even if Ingres were a drop-in, I can't imagine a CA executive deciding to start out with an unproven SAP/Ingres combination, given the dire consequences of not having it working in time. And, from what little I know of it, *nothing* is a drop-in where SAP is concerned. I don't doubt for a minute that we are talking with SAP, but I'm not privy to that sort of thing and have no juicy rumors to feed anyone. Alas. Karl ...

Re: [Info-ingres] Re: What animal should Ingres be? #2
Hi there, "Roy Hann" <specially@processed.almost.meat> wrote: >"Emiliano" <Emiliano@Iris-Advies.nl> wrote in message >news:d395$43973fce$3ec29f9d$4203@news.chello.nl... >> Tonyd08068@netscape.net wrote: >> >> > If we're going to go to all that hassle, why not take a leaf out of the >PostgreSQL book and > >> I don't really care whether it's java, or python, or C#, or Tcl or even >> lisp; it's just that many of the developers around me that have grown up >> on the 'regular' procedural and/or OO languages find Ingres' SP >> language arcane. > >It is certainly feeble, and it lacks a debugger, or anything that would >allow you to impose any kind of release discipline, but to say it is arcane >is to give it more credit than it is due. For industrial use, the procedure language we have just now is very definitely, shall we say, "under featured". I wouldn't say the language is syntactically arcane though. What can be rather arcane is the run-time behaviour of some language features, and that's where the lack of useful development tools bites, big style. >(Unfortunately I would very much >prefer to see a purely set-oriented language, but I think that would just do >their little heads in completely. I think I might be in a minority of about >two with this, and even two is assuming "M.T."...

Re: [Info-ingres] Re: What animal should Ingres be? #2
Hi there, Michael Leo <mleo@cariboulake.com> wrote: >Sorry. I lied. Not my last word. The missile missed. Packer >fans are happy in heaven. > Given the Packers' performances this season ... >Roy Hann wrote: > >>"Michael Leo" <mleo@cariboulake.com> wrote in message >>news:mailman.1134079081.15832.info-ingres@cariboulake.com... >> >> >> [snip] >> >>I am sure that happened, but my recollection of the same era is that it was >>the lack of row-level locking that got everyone all twisted. Same argument >>though, I agree. But I also think this makes the mistake of assuming that >>what people say and what they think are in any way related. My cynical view >>is that a lot of people wanted to use Oracle back then only because it was >>"cool", and then went looking for differences that could be made to look >>like reasons. Very few people "need" ALTER TABLE and very few people can >>effectively use row-level locking. Adding them hasn't helped because that >>was never the real problem. Lack of "cool" was the real problem. Cool >>comes from being endorsed by the right customers not from feature X or Y. >> >> >Boy I missed the boat here. Not only row-level locking, but >Oracle's row-versioning. I am the DBA of several Ingres, SQL Server, >and Or...

RE: [Info-ingres] Re: How Different is Ingres? #2
This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_001_01C5820F.3450AD70 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit > -----Original Message----- > From: David Richard [mailto:Richard.David@aah.co.uk] > Sent: 06 July 2005 09:49 > To: info-ingres@cariboulake.com > Subject: RE: [Info-ingres] Re: How Different is Ingres? > > Postgres' Multi Version Concurrency Control (MVCC) is explained here http://www.developer.com/open/article.php/877181 and here > http://www.onlamp.com/pub/a/onlamp/2001/05/25/postgresql_mvcc.html. > I can't see why a similar locking option couldn't be added to Ingres in time?? > MVCC _may_ provide greater concurrency, but its far from perfect. > This old article compares DBMS locking methods, traps and pitfalls:- http://www.dbpd.com/vault/saracco.htm Forgot to add the following link to illustrate what can go wrong using MVCC:- http://asktom.oracle.com/~tkyte/wc.html ************************************************************************ DISCLAIMER The information contained in this e-mail is confidential and is intended for the recipient only. If you have received it in error, please notify us immediately by reply e-mail and then delete it from your system. Please do not copy it or use it for any other purposes, or disclose the ...

Re: [Info-ingres] Re: What animal should Ingres be? #2
Hi Mike, Michael Leo <mleo@cariboulake.com> wrote: [ lots of very true stuff about Java snipped ] A lot of what you say about Java is very true; in our shop, we have Java bytecode bouncing between Windows, Solaris, AIX and Linux. Any issues we've bumped into have been JVM implementation issues, not problems with the bytecode itself. And a JVM might open up some other interesting possibilities for sickos like me - JIProlog, anybody, or maybe even ML/J ? However, for the task at hand, my main gripes can be enumerated as : - the core Java language is weak (generally, and specifically for database applications) - if we implement an "extended, safe subset" of Java, the core language is pretty much all we'll be left with; many of those class libraries and frameworks won't be available - the interpreter for the database procedure language will be built in to the database server; therefore it will have to run on every platform the database server runs on Points 2 and 3 effectively negate much of the good stuff that allows me to ignore point 1 when using Java for applications outside the DBMS server. Don't get me wrong; while I honestly don't see what the fuss is with Java as either a language or technology (everything I like about Java, such as worthwhile libraries, machine independent portable bytecodes, has been available for decades - BCPL & Ocode, say - and everything I don't much like about Java, namely the whole...

Re: [Info-ingres] Re: What animal should Ingres be? #2
Hi Mike, Michael Leo <mleo@cariboulake.com> wrote: [ snippage ] >Tony, > >You inspired me to look at how PostgreSQL does it. >I'm liking what I see, but I can't seem to find anything about >a Java binding. > The two I had a brief look at were PL/Java - http://gborg.postgresql.org/project/pljava/projdisplay.php PL-J - http://plj.codehaus.org Both seem to be reasonably current, so there is very probably some horrible, wrinkly difference between the two - otherwise they wouldn't have any justification for hanging around. PL/Java seems to make a thing about SQL2003 compatibility (was that fairly directly influenced by SQLJ ?). [ snips ] >(If anyone chimes in about CORBA, I will puke. Find me ANY two >CORBA clients from the SAME vendor that can work together and >I will eat my entire 6.4 documentation set including the binders.) > You don't wanna go there - either CORBA or chewing a 6.4 doc stack. (The question raises itself - *which* 6.4 documentation set ? We got several, and no two of the neat holders contained exactly the same set of books !) [ snips ] > >Cheers, > >Mikey > Cheers, - Tony __________________________________________________________________ Switch to Netscape Internet Service. As low as $9.95 a month -- Sign up today at http://isp.netscape.com/register Netscape. Just the Net You Need. New! Netscape Toolbar for Internet Explorer Se...

Web resources about - RE: NTP ( Network Time protocol ) With Ingres #2 - comp.databases.ingres

Kyoto Protocol - Wikipedia, the free encyclopedia
... in their carbon dioxide emissions from fuel combustion between 1990 and 2009. For more detailed country/region information, see Kyoto Protocol ...

Jarryd Hayne's return to San Francisco 49ers line-up may hinge on NFL's strict concussion protocol
After suffering a concussion in the disastrous loss to Cleveland, the man standing in Jarryd Hayne's way could find he has no choice but to sit ...

Common payment processing protocols found to be full of flaws
Stealing PINs and pillaging bank accounts are both trivial.

NFL Considering Changes to Officiating Protocol for Playoffs - Bleacher Report
With the NFL officials under the microscope more than ever, the league is contemplating making changes to officiating protocol for the postseason... ...


Theme parks add extra safety protocols - Videos - CBS News
Major U.S. theme parks across the country, including Disney, SeaWorld and Universal Studios, are tightening security measures at entry gates. ...


IDG Contributor Network: OAuth 2.0: The protocol at the center of the universe
I am writing this article on the iPad Mini using the Editorial app. This app is connected to my Dropbox account and automatically synchronizes ...

'Routine,' 'standard,' and 'protocol' for police usually means overkill for black folks
A neighbor sees someone who may be breaking into a home. Civic-minded and filled with neighborly concern, the witness dials 911 and asks police ...

Imran promises to end VIP protocol in KP
... Imran Khan has said that his party’s government in Khyber Pakhtunkhwa will take measures to end the practice of halting traffic for VIP protocol. ...

Resources last updated: 2/22/2016 2:30:42 AM