Performance Issues

I am running IDS 9.21.HC6 on HP UX 11.0. Until recently everything has been running smothly but over the last 2/3 weeks users have reported problems with one of our applications delivered over the web. I have also noticed that the % reads cached figure (onstat -p) is very low, sometimes 0%, so I knew something was very wrong !! Last week I noticed multiple ETXFilterServers running (for the Excalibur index in the database). I killed off the excess and bounced the server. And all seemed OK. The % reads cached figure was up in the 70's, not as good as it should be but better. At the weekend I was running lots of dbexports/dbimports and the Êched figure was about 85%, as expected. As soon as the users logged on this morning and started running our web applications the Êched figure slowly dropped and is now 19.44%. I have 40000 buffers defined for 8 LRU queues, onstat -b shows 836 modified, 0 resident, 40000 total onstat -F shows 2 Fg write, 534203 LRU writes, 57178 Chunk writes onstat -R shows 5 dirty, 40000 queued, 40000 total, start clean at 4% (160), stop at 2% As far as I am aware nothing has changed on the Database server but I am not sure about the Web Server. I am also not very familiar with the Excalibur Text Search Datablade and the ETXFilterServers. Our web applicaiton is a document management system delivered over the web and users are experiencing problems bookings documents back in, as it seems to ages and sometimes the browser is timing out (we currently have a 5 minute timeout set on the server). Does anyone have any ideas what might be causing these problems, or anything else I need to check/monitor ??? Sue

Sue 40,000 buffers seems a bit low, especially as you have hit some foreground writes. What does a full onstat -p show (especially over a period of time, what is changing the most). How long are checkpoints taking? If you have the memory available I would try doubling buffers as a first step and are fully monitor checkpoints with a view to reducing LRU percentages. Probably need to double (or more) number of LRUs and CLEANERs. Any clues from Unix side? sar -d ? sar ? Keith -> -----Original Message----- -> From: SUE SIMMONDS [mailto:sues@hmgcc.gsi.gov.uk] -> Sent: Friday, May 13, 2005 10:33 PM -> To: ids@iiug.org -> Subject: Performance Issues [4950] -> -> -> I am running IDS 9.21.HC6 on HP UX 11.0. Until recently -> everything has been running smothly but over the last 2/3 -> weeks users have reported problems with one of our -> applications delivered over the web. -> -> I have also noticed that the % reads cached figure (onstat -> -p) is very low, sometimes 0%, so I knew something was very wrong !! -> -> Last week I noticed multiple ETXFilterServers running (for -> the Excalibur index in the database). I killed off the -> excess and bounced the server. And all seemed OK. The % -> reads cached figure was up in the 70's, not as good as it -> should be but better. -> -> At the weekend I was running lots of dbexports/dbimports and -> the Êched figure was about 85%, as expected. As soon as the -> users logged on this morning and started running our web -> applications the Êched figure slowly dropped and is now 19.44%. -> -> I have 40000 buffers defined for 8 LRU queues, -> -> onstat -b shows 836 modified, 0 resident, 40000 total -> -> onstat -F shows 2 Fg write, 534203 LRU writes, 57178 Chunk writes -> -> onstat -R shows 5 dirty, 40000 queued, 40000 total, start clean at 4% -> (160), stop at 2% -> -> As far as I am aware nothing has changed on the Database -> server but I am not sure about the Web Server. I am also not -> very familiar with the Excalibur Text Search Datablade and -> the ETXFilterServers. -> -> Our web applicaiton is a document management system -> delivered over the web and users are experiencing problems -> bookings documents back in, as it seems to ages and -> sometimes the browser is timing out (we currently have a 5 -> minute timeout set on the server). -> -> Does anyone have any ideas what might be causing these -> problems, or anything else I need to check/monitor ??? -> -> Sue -> ******************************************************************************** ** This message is sent in strict confidence for the addressee only. It may contain legally privileged information. The contents are not to be disclosed to anyone other than the addressee. Unauthorised recipients are requested to preserve this confidentiality and to advise the sender immediately of any error in transmission. This footnote also confirms that this email message has been swept for the presence of computer viruses, however we cannot guarantee that this message is free from such problems. ******************************************************************************** **

Sue Êched reads now looks OK, don't believe the cached writes ! ,I think the engine is lying. IDS 9.21 is very old, any chance of updating to 9.4? The X in HC6X6 indicates a patched engine, even more reason to upgrade to a more supported version. What are your checkpoints like? What about onstat -F? I there any user perception of slow running? Keith -----Original Message----- From: sues@hmgcc.gsi.gov.uk [mailto:sues@hmgcc.gsi.gov.uk] Sent: Monday, May 16, 2005 2:10 PM To: Simmons, Keith; ids@iiug.org Subject: Re: Performance Issues [4950] I have now increased the buffers from 40000 to 80000, LRU from 10 to 20 and CLEANERS from 12 to 24 and bounced the server. The server has been up for just over 1 hour and onstat -p shows Informix Dynamic Server 2000 Version 9.21.HC6X6 -- On-Line -- Up 01:22:18 -- 377200 Kbytes Profile dskreads pagreads bufreads Êched dskwrits pagwrits bufwrits Êched 96659 125354 23467579 99.59 151039 104045 101784 0.00 isamtot open start read write rewrite delete commit rollbk 24314753 800197 1291125 15254883 3177 232 2364 588 3 gp_read gp_write gp_rewrt gp_del gp_alloc gp_free gp_curs 24609 12 2271 8 0 0 22 ovlock ovuserthread ovbuff usercpu syscpu numckpts flushes 0 0 0 1004.44 50.46 17 36 bufwaits lokwaits lockreqs deadlks dltouts ckpwaits compress seqscans 3239 28 25717535 1 0 5 491 12376 ixda-RA idx-RA da-RA RA-pgsused lchwaits 4422 136 32039 36480 71 Sue ----- Original Message ----- From: Simmons, Keith To: SUE SIMMONDS ; ids@iiug.org Sent: 16 May 2005 9:21 AM Subject: RE: Performance Issues [4950] Sue 40,000 buffers seems a bit low, especially as you have hit some foreground writes. What does a full onstat -p show (especially over a period of time, what is changing the most). How long are checkpoints taking? If you have the memory available I would try doubling buffers as a first step and are fully monitor checkpoints with a view to reducing LRU percentages. Probably need to double (or more) number of LRUs and CLEANERs. Any clues from Unix side? sar -d ? sar ? Keith -> -----Original Message----- -> From: SUE SIMMONDS [mailto:sues@hmgcc.gsi.gov.uk] -> Sent: Friday, May 13, 2005 10:33 PM -> To: ids@iiug.org -> Subject: Performance Issues [4950] -> -> -> I am running IDS 9.21.HC6 on HP UX 11.0. Until recently -> everything has been running smothly but over the last 2/3 -> weeks users have reported problems with one of our -> applications delivered over the web. -> -> I have also noticed that the % reads cached figure (onstat -> -p) is very low, sometimes 0%, so I knew something was very wrong !! -> -> Last week I noticed multiple ETXFilterServers running (for -> the Excalibur index in the database). I killed off the -> excess and bounced the server. And all seemed OK. The % -> reads cached figure was up in the 70's, not as good as it -> should be but better. -> -> At the weekend I was running lots of dbexports/dbimports and -> the Êched figure was about 85%, as expected. As soon as the -> users logged on this morning and started running our web -> applications the Êched figure slowly dropped and is now 19.44%. -> -> I have 40000 buffers defined for 8 LRU queues, -> -> onstat -b shows 836 modified, 0 resident, 40000 total -> -> onstat -F shows 2 Fg write, 534203 LRU writes, 57178 Chunk writes -> -> onstat -R shows 5 dirty, 40000 queued, 40000 total, start clean at 4% -> (160), stop at 2% -> -> As far as I am aware nothing has changed on the Database -> server but I am not sure about the Web Server. I am also not -> very familiar with the Excalibur Text Search Datablade and -> the ETXFilterServers. -> -> Our web applicaiton is a document management system -> delivered over the web and users are experiencing problems -> bookings documents back in, as it seems to ages and -> sometimes the browser is timing out (we currently have a 5 -> minute timeout set on the server). -> -> Does anyone have any ideas what might be causing these -> problems, or anything else I need to check/monitor ??? -> -> Sue -> ******************************************************************************** ** This message is sent in strict confidence for the addressee only. It may contain legally privileged information. The contents are not to be disclosed to anyone other than the addressee. Unauthorised recipients are requested to preserve this confidentiality and to advise the sender immediately of any error in transmission. This footnote also confirms that this email message has been swept for the presence of computer viruses, however we cannot guarantee that this message is free from such problems. ******************************************************************************** ** PLEASE NOTE: THE ABOVE MESSAGE WAS RECEIVED FROM THE INTERNET. On entering the GSi, this email was scanned for viruses by the Government Secure Intranet (GSi) virus scanning service supplied exclusively by Energis in partnership with MessageLabs. Please see http://www.gsi.gov.uk/main/notices/information/gsi-003-2002.pdf for further details. In case of problems, please call your organisational IT helpdesk The information contained in this message (and any attachments) may be confidential and is intended for the sole use of the named addressee. Access, copying, alteration or re-use of the e-mail by anyone other than the intended recipient is unauthorised. If you are not the intended recipient please advise the sender immediately by returning the e-mail and deleting it from your system. This information may be exempt from disclosure under Freedom Of Information Act 2000 and may be subject to exemption under other UK information legislation. Refer disclosure requests to the Information Officer The original of this email was scanned for viruses by the Government Secure Intranet (GSi) virus scanning service supplied exclusively by Energis in partnership with MessageLabs. On leaving the GSi this email was certified virus-free

Sue on_archive is not supplied as it is unreliable, get off it NOW, even = ontape is better, more reliable, simple to use etc. etc. Have you run 'update statistics' recently? =20 Keith -----Original Message----- From: sues@hmgcc.gsi.gov.uk [mailto:sues@hmgcc.gsi.gov.uk] Sent: Tuesday, May 17, 2005 10:42 AM To: Simmons, Keith; ids@iiug.org Subject: Re: Performance Issues [4950]=20 =20 If I reset the stats (onstat -z) the % writes cached figure starts = initially at about=20 onstat -p is now showing=20 Informix Dynamic Server 2000 Version 9.21.HC6X6 -- On-Line -- Up = 21:46:09 -- 377200 Kbytes =20 Profile dskreads pagreads bufreads Êched dskwrits pagwrits bufwrits Êched 1599258 7628062 139203779 98.85 1297984 773787 1091868 0.00 =20 =20 I know we are running an old patched version of the engine but we are = currently running on-archive for our backups and on-archive is not = supported, or even supplied with 9.4. I am currently trying to upgrade = to on-bar before upgrading to 9.4, but that is another issue !!! =20 Fuzzy Checkpoints are generally taking between 1 and 3 seconds but = occassionally as long as 30 !!! =20 onstat -F shows =20 Fg Writes LRU Writes Chunk Writes=20 0 500148 108425 =20 =20 And yes the users have been complaining, things are obviously running = slower than they used to. =20 I also need to check the fragmentation of the main table in my document = management system, as suggested by Robert Roussey in a previous posting. =20 =20 Sue =20 ----- Original Message -----=20 From: Simmons, Keith <mailto:keith.simmons@office2office.biz> =20 To: sues@hmgcc.gsi.gov.uk ; ids@iiug.org=20 Sent: 17 May 2005 9:12 AM Subject: RE: Performance Issues [4950]=20 Sue Êched reads now looks OK, don't believe the cached writes ! ,I think = the engine is lying. IDS 9.21 is very old, any chance of updating to = 9.4? The X in HC6X6 indicates a patched engine, even more reason to = upgrade to a more supported version. What are your checkpoints like? What about onstat -F? I there any user = perception of slow running? Keith -----Original Message----- From: sues@hmgcc.gsi.gov.uk [mailto:sues@hmgcc.gsi.gov.uk] Sent: Monday, May 16, 2005 2:10 PM To: Simmons, Keith; ids@iiug.org Subject: Re: Performance Issues [4950]=20 I have now increased the buffers from 40000 to 80000, LRU from 10 to 20 = and CLEANERS from 12 to 24 and bounced the server.=20 The server has been up for just over 1 hour and onstat -p shows Informix Dynamic Server 2000 Version 9.21.HC6X6 -- On-Line -- Up = 01:22:18 -- 377200 Kbytes Profile dskreads pagreads bufreads Êched dskwrits pagwrits bufwrits Êched 96659 125354 23467579 99.59 151039 104045 101784 0.00 =20 isamtot open start read write rewrite delete commit = rollbk 24314753 800197 1291125 15254883 3177 232 2364 588 = 3 gp_read gp_write gp_rewrt gp_del gp_alloc gp_free gp_curs=20 24609 12 2271 8 0 0 22 =20 ovlock ovuserthread ovbuff usercpu syscpu numckpts flushes=20 0 0 0 1004.44 50.46 17 36 =20 bufwaits lokwaits lockreqs deadlks dltouts ckpwaits compress seqscans 3239 28 25717535 1 0 5 491 12376 =20 ixda-RA idx-RA da-RA RA-pgsused lchwaits 4422 136 32039 36480 71 =20 Sue ----- Original Message -----=20 From: Simmons, Keith=20 To: SUE SIMMONDS ; ids@iiug.org=20 Sent: 16 May 2005 9:21 AM Subject: RE: Performance Issues [4950]=20 Sue 40,000 buffers seems a bit low, especially as you have hit some foreground writes. What does a full onstat -p show (especially over a period of time, what is changing the most). How long are checkpoints taking? If you have the memory available I would try doubling buffers as a first step and are fully monitor checkpoints with a view to reducing LRU percentages. Probably need to double (or more) number of LRUs and CLEANERs. Any clues from Unix side? sar -d ? sar ? Keith -> -----Original Message----- -> From: SUE SIMMONDS [mailto:sues@hmgcc.gsi.gov.uk] -> Sent: Friday, May 13, 2005 10:33 PM -> To: ids@iiug.org -> Subject: Performance Issues [4950]=20 ->=20 ->=20 -> I am running IDS 9.21.HC6 on HP UX 11.0. Until recently=20 -> everything has been running smothly but over the last 2/3=20 -> weeks users have reported problems with one of our=20 -> applications delivered over the web. ->=20 -> I have also noticed that the % reads cached figure (onstat=20 -> -p) is very low, sometimes 0%, so I knew something was very wrong !! ->=20 -> Last week I noticed multiple ETXFilterServers running (for=20 -> the Excalibur index in the database). I killed off the=20 -> excess and bounced the server. And all seemed OK. The %=20 -> reads cached figure was up in the 70's, not as good as it=20 -> should be but better. ->=20 -> At the weekend I was running lots of dbexports/dbimports and=20 -> the =CAched figure was about 85%, as expected. As soon as the=20 -> users logged on this morning and started running our web=20 -> applications the =CAched figure slowly dropped and is now 19.44%. ->=20 -> I have 40000 buffers defined for 8 LRU queues, ->=20 -> onstat -b shows 836 modified, 0 resident, 40000 total ->=20 -> onstat -F shows 2 Fg write, 534203 LRU writes, 57178 Chunk writes ->=20 -> onstat -R shows 5 dirty, 40000 queued, 40000 total, start clean at 4% -> (160), stop at 2% ->=20 -> As far as I am aware nothing has changed on the Database=20 -> server but I am not sure about the Web Server. I am also not=20 -> very familiar with the Excalibur Text Search Datablade and=20 -> the ETXFilterServers. ->=20 -> Our web applicaiton is a document management system=20 -> delivered over the web and users are experiencing problems=20 -> bookings documents back in, as it seems to ages and=20 -> sometimes the browser is timing out (we currently have a 5=20 -> minute timeout set on the server). ->=20 -> Does anyone have any ideas what might be causing these=20 -> problems, or anything else I need to check/monitor ??? ->=20 -> Sue=20 ->=20 *************************************************************************= ********* This message is sent in strict confidence for the addressee only. It = may contain legally privileged information. The contents are not to be = disclosed to anyone other than the addressee. Unauthorised recipients are = requested to preserve this confidentiality and to advise the sender immediately of = any error in transmission. This footnote also confirms that this email message has been swept for = the presence of computer viruses, however we cannot guarantee that this = message is free from such problems. *************************************************************************= ********* PLEASE NOTE: THE ABOVE MESSAGE WAS RECEIVED FROM THE INTERNET. =20 On entering the GSi, this email was scanned for viruses by the = Government Secure Intranet (GSi) virus scanning service supplied = exclusively by Energis in partnership with MessageLabs. =20 Please see = http://www.gsi.gov.uk/main/notices/information/gsi-003-2002.pdf for =

Beginning to get stuck. :-( Is the underlying Unix Server performing OK? N= o contention/bottlenecks on disk, memory or CPU? I assume nothing specific= changed (application wise) about three weeks ago that could have caused th= ese problems? Have you monitored onstat -u to see if anyone is making exces= sive reads (and trapping the sql (onstst -g sql 'sesid')) =20 Keith -----Original Message----- From: sues@hmgcc.gsi.gov.uk [mailto:sues@hmgcc.gsi.gov.uk] Sent: Tuesday, May 17, 2005 12:22 PM To: Simmons, Keith; ids@iiug.org Subject: Re: Performance Issues [4950]=20 =20 On-Archive has been totally reliable and works well for us but I am despera= tely trying to move to on-bar. Update statistics is run on all databases every night.=20 =20 Sue ----- Original Message -----=20 From: Simmons, Keith <mailto:keith.simmons@office2office.biz> =20 To: sues@hmgcc.gsi.gov.uk ; ids@iiug.org=20 Sent: 17 May 2005 10:54 AM Subject: RE: Performance Issues [4950]=20 Sue on_archive is not supplied as it is unreliable, get off it NOW, even ontape= is better, more reliable, simple to use etc. etc. Have you run 'update statistics' recently? =20 Keith -----Original Message----- From: sues@hmgcc.gsi.gov.uk [mailto:sues@hmgcc.gsi.gov.uk] Sent: Tuesday, May 17, 2005 10:42 AM To: Simmons, Keith; ids@iiug.org Subject: Re: Performance Issues [4950]=20 =20 If I reset the stats (onstat -z) the % writes cached figure starts initiall= y at about=20 onstat -p is now showing=20 Informix Dynamic Server 2000 Version 9.21.HC6X6 -- On-Line -- Up 21:46:09= -- 377200 Kbytes =20 Profile dskreads pagreads bufreads Êched dskwrits pagwrits bufwrits Êched 1599258 7628062 139203779 98.85 1297984 773787 1091868 0.00 =20 =20 I know we are running an old patched version of the engine but we are curre= ntly running on-archive for our backups and on-archive is not supported, or= even supplied with 9.4. I am currently trying to upgrade to on-bar before = upgrading to 9.4, but that is another issue !!! =20 Fuzzy Checkpoints are generally taking between 1 and 3 seconds but occassio= nally as long as 30 !!! =20 onstat -F shows =20 Fg Writes LRU Writes Chunk Writes=20 0 500148 108425 =20 =20 And yes the users have been complaining, things are obviously running slowe= r than they used to. =20 I also need to check the fragmentation of the main table in my document man= agement system, as suggested by Robert Roussey in a previous posting. =20 =20 Sue =20 ----- Original Message -----=20 From: Simmons, Keith <mailto:keith.simmons@office2office.biz> =20 To: sues@hmgcc.gsi.gov.uk ; ids@iiug.org=20 Sent: 17 May 2005 9:12 AM Subject: RE: Performance Issues [4950]=20 Sue Êched reads now looks OK, don't believe the cached writes ! ,I think the = engine is lying. IDS 9.21 is very old, any chance of updating to 9.4? The X= in HC6X6 indicates a patched engine, even more reason to upgrade to a more= supported version. What are your checkpoints like? What about onstat -F? I there any user perc= eption of slow running? Keith -----Original Message----- From: sues@hmgcc.gsi.gov.uk [mailto:sues@hmgcc.gsi.gov.uk] Sent: Monday, May 16, 2005 2:10 PM To: Simmons, Keith; ids@iiug.org Subject: Re: Performance Issues [4950]=20 I have now increased the buffers from 40000 to 80000, LRU from 10 to 20 and= CLEANERS from 12 to 24 and bounced the server.=20 The server has been up for just over 1 hour and onstat -p shows Informix Dynamic Server 2000 Version 9.21.HC6X6 -- On-Line -- Up 01:22:18= -- 377200 Kbytes Profile dskreads pagreads bufreads Êched dskwrits pagwrits bufwrits Êched 96659 125354 23467579 99.59 151039 104045 101784 0.00 =20 isamtot open start read write rewrite delete commit rol= lbk 24314753 800197 1291125 15254883 3177 232 2364 588 3 gp_read gp_write gp_rewrt gp_del gp_alloc gp_free gp_curs=20 24609 12 2271 8 0 0 22 =20 ovlock ovuserthread ovbuff usercpu syscpu numckpts flushes=20 0 0 0 1004.44 50.46 17 36 =20 bufwaits lokwaits lockreqs deadlks dltouts ckpwaits compress seqscans 3239 28 25717535 1 0 5 491 12376 =20 ixda-RA idx-RA da-RA RA-pgsused lchwaits 4422 136 32039 36480 71 =20 Sue ----- Original Message -----=20 From: Simmons, Keith=20 To: SUE SIMMONDS ; ids@iiug.org=20 Sent: 16 May 2005 9:21 AM Subject: RE: Performance Issues [4950]=20 Sue 40,000 buffers seems a bit low, especially as you have hit some foreground writes. What does a full onstat -p show (especially over a period of time, what is changing the most). How long are checkpoints taking? If you have the memory available I would try doubling buffers as a first step and are fully monitor checkpoints with a view to reducing LRU percentages. Probably need to double (or more) number of LRUs and CLEANERs. Any clues from Unix side? sar -d ? sar ? Keith -> -----Original Message----- -> From: SUE SIMMONDS [mailto:sues@hmgcc.gsi.gov.uk] -> Sent: Friday, May 13, 2005 10:33 PM -> To: ids@iiug.org -> Subject: Performance Issues [4950]=20 ->=20 ->=20 -> I am running IDS 9.21.HC6 on HP UX 11.0. Until recently=20 -> everything has been running smothly but over the last 2/3=20 -> weeks users have reported problems with one of our=20 -> applications delivered over the web. ->=20 -> I have also noticed that the % reads cached figure (onstat=20 -> -p) is very low, sometimes 0%, so I knew something was very wrong !! ->=20 -> Last week I noticed multiple ETXFilterServers running (for=20 -> the Excalibur index in the database). I killed off the=20 -> excess and bounced the server. And all seemed OK. The %=20 -> reads cached figure was up in the 70's, not as good as it=20 -> should be but better. ->=20 -> At the weekend I was running lots of dbexports/dbimports and=20 -> the =CAched figure was about 85%, as expected. As soon as the=20 -> users logged on this morning and started running our web=20 -> applications the =CAched figure slowly dropped and is now 19.44%. ->=20 -> I have 40000 buffers defined for 8 LRU queues, ->=20 -> onstat -b shows 836 modified, 0 resident, 40000 total ->=20 -> onstat -F shows 2 Fg write, 534203 LRU writes, 57178 Chunk writes ->=20 -> onstat -R shows 5 dirty, 40000 queued, 40000 total, start clean at 4% -> (160), stop at 2% ->=20 -> As far as I am aware nothing has changed on the Database=20 -> server but I am not sure about the Web Server. I am also not=20 -> very familiar with the Excalibur Text Search Datablade and=20 -> the ETXFilterServers. ->=20 -> Our web applicaiton is a document management system=20 -> delivered over the web and users are experiencing problems=20 -> bookings documents back in, as it seems to ages and=20 -> sometimes the browser is timing out (we currently have a 5=20 -> minute timeout set on the server). ->=20 -> Does anyone have any ideas what might be causing these=20 -> problems, or anything else I need to check/monitor ??? ->=20 -> Sue=20 ->=20 [Keith Simmons] <Snip>=20 ***********************************

Sue, I'm a bit concerned over the number of Lock Requests you have. 25.7 million in just over an hour? That's 35000+ per minute! Having 1 deadlock and 5 timeouts already definitely shows some sort of SQL coding issues. Other evidence of SQL issues is the lack of Buffered Writes. The Êched seems to be correct (if I recall, (BufWrts - DskWrts) / BufWrts) which comes out *negative*. I'd look at increasing your Buffers again, and monitor when the longer checkpoints occur. Are the users doing any type of 'batch' insert? Could a program be running a large Insert or Update cursor? If yes, maybe tuning the SQL could improve the performance. If not, decrease your LRU_Max and LRU_Min values. This will increase the LRU writes, but should help with the long checkpoints. Michael Hoffman > > Beginning to get stuck. :-( Is the underlying Unix Server performing OK? N= > o contention/bottlenecks on disk, memory or CPU? I assume nothing specific= > changed (application wise) about three weeks ago that could have caused th= > ese problems? Have you monitored onstat -u to see if anyone is making exces= > sive reads (and trapping the sql (onstst -g sql 'sesid')) > =20 > Keith > > -----Original Message----- > From: sues@hmgcc.gsi.gov.uk [mailto:sues@hmgcc.gsi.gov.uk] > Sent: Tuesday, May 17, 2005 12:22 PM > To: Simmons, Keith; ids@iiug.org > Subject: Re: Performance Issues [4950]=20 > > > =20 > On-Archive has been totally reliable and works well for us but I am despera= > tely trying to move to on-bar. > Update statistics is run on all databases every night.> =20 > =20 > Sue > > > ----- Original Message -----=20 > From: Simmons, Keith <mailto:keith.simmons@office2office.biz> =20 > To: sues@hmgcc.gsi.gov.uk ; ids@iiug.org=20 > Sent: 17 May 2005 10:54 AM > Subject: RE: Performance Issues [4950]=20 > > Sue > on_archive is not supplied as it is unreliable, get off it NOW, even ontape= > is better, more reliable, simple to use etc. etc. > Have you run 'update statistics' recently? > =20 > Keith > > -----Original Message----- > From: sues@hmgcc.gsi.gov.uk [mailto:sues@hmgcc.gsi.gov.uk] > Sent: Tuesday, May 17, 2005 10:42 AM > To: Simmons, Keith; ids@iiug.org > Subject: Re: Performance Issues [4950]=20 > > > =20 > If I reset the stats (onstat -z) the % writes cached figure starts initiall= > y at about=20 > onstat -p is now showing=20> > Informix Dynamic Server 2000 Version 9.21.HC6X6 -- On-Line -- Up 21:46:09= > -- 377200 Kbytes > =20 > Profile > dskreads pagreads bufreads Êched dskwrits pagwrits bufwrits Êched > 1599258 7628062 139203779 98.85 1297984 773787 1091868 0.00 =20 > =20 > I know we are running an old patched version of the engine but we are curre= > ntly running on-archive for our backups and on-archive is not supported, or= > even supplied with 9.4. I am currently trying to upgrade to on-bar before = > upgrading to 9.4, but that is another issue !!! > =20 > Fuzzy Checkpoints are generally taking between 1 and 3 seconds but occassio= > nally as long as 30 !!! > =20 > onstat -F shows > =20> > Fg Writes LRU Writes Chunk Writes=20 > 0 500148 108425 =20 > =20 > And yes the users have been complaining, things are obviously running slowe= > r than they used to. > =20 > I also need to check the fragmentation of the main table in my document man= > agement system, as suggested by Robert Roussey in a previous posting. > =20 > =20 > Sue > =20 > > > > ----- Original Message -----=20 > From: Simmons, Keith <mailto:keith.simmons@office2office.biz> =20 > To: sues@hmgcc.gsi.gov.uk ; ids@iiug.org=20 > Sent: 17 May 2005 9:12 AM > Subject: RE: Performance Issues [4950]=20 > > > Sue > > Êched reads now looks OK, don't believe the cached writes ! ,I think the = > engine is lying. IDS 9.21 is very old, any chance of updating to 9.4? The X= > in HC6X6 indicates a patched engine, even more reason to upgrade to a more= > supported version. > What are your checkpoints like? What about onstat -F? I there any user perc= > eption of slow running? > > Keith > > -----Original Message----- > From: sues@hmgcc.gsi.gov.uk [mailto:sues@hmgcc.gsi.gov.uk] > Sent: Monday, May 16, 2005 2:10 PM > To: Simmons, Keith; ids@iiug.org > Subject: Re: Performance Issues [4950]=20 > > I have now increased the buffers from 40000 to 80000, LRU from 10 to 20 and= > CLEANERS from 12 to 24 and bounced the server.=20 > The server has been up for just over 1 hour and onstat -p shows > > Informix Dynamic Server 2000 Version 9.21.HC6X6 -- On-Line -- Up 01:22:18= > -- 377200 Kbytes > > Profile > dskreads pagreads bufreads Êched dskwrits pagwrits bufwrits Êched > 96659 125354 23467579 99.59 151039 104045 101784 0.00 =20 > > isamtot open start read write rewrite delete commit rol= > lbk > 24314753 800197 1291125 15254883 3177 232 2364 588 3 > > gp_read gp_write gp_rewrt gp_del gp_alloc gp_free gp_curs=20 > 24609 12 2271 8 0 0 22 =20 > > ovlock ovuserthread ovbuff usercpu syscpu numckpts flushes=20 > 0 0 0 1004.44 50.46 17 36 =20 > > bufwaits lokwaits lockreqs deadlks dltouts ckpwaits compress seqscans > 3239 28 25717535 1 0 5 491 12376 =20 > > ixda-RA idx-RA da-RA RA-pgsused lchwaits > 4422 136 32039 36480 71 =20 > > Sue > > ----- Original Message -----=20 > From: Simmons, Keith=20 > To: SUE SIMMONDS ; ids@iiug.org=20 > Sent: 16 May 2005 9:21 AM > Subject: RE: Performance Issues [4950]=20 > > > Sue > > 40,000 buffers seems a bit low, especially as you have hit > some foreground writes. What does a full onstat -p show (especially > over a period of time, what is changing the most). How long are > checkpoints taking? > If you have the memory available I would try doubling buffers as a > first step and are fully monitor checkpoints with a view to reducing > LRU percentages. Probably need to double (or more) number of LRUs > and CLEANERs. > Any clues from Unix side? sar -d ? sar ? > > Keith > > > -> -----Original Message----- > -> From: SUE SIMMONDS [mailto:sues@hmgcc.gsi.gov.uk] > -> Sent: Friday, May 13, 2005 10:33 PM > -> To: ids@iiug.org > -> Subject: Performance Issues [4950]=20 > ->=20 > ->=20 > -> I am running IDS 9.21.HC6 on HP UX 11.0. Until recently=20 > -> everything has been running smothly but over the last 2/3=20 > -> weeks users have reported problems with one of our=20 > -> applications delivered over the web. > ->=20 > -> I have also noticed that the % reads cached figure (onstat=20 > -> -p) is very low, sometimes 0%, so I knew something was very wrong !! > ->=20 > -> Last week I noticed multiple ETXFilterServers running (for=20 > -> the Excalibur index in the database). I killed off the=20 > -> excess and bounced the server. And all seemed OK. The %=20 > -> reads cached figure was up in the 70's, not as good as it=20 > -> should be but better. > ->=20 > -> At the weekend I was running lots of dbexports/dbimports and=20 > -> the =CAched figure was about 85%, as expected. As soon as the=20 > -> users logged on this morning and started ru