Re: IDS2000 Aborting Long TX problem

Log 6 is larger than the rest of your logs combined. So if log 6 gets full then you hit your LTXHWM and LTXEHWM . You really should have all logs the exact same size. Jo ----- Original Message ----- From: "Sean Kelsey" <seank@marketinfoserv.com> To: <informix-list@iiug.org> Sent: Wednesday, November 22, 2000 11:39 AM Subject: IDS2000 Aborting Long TX problem > We are getting some serious problems with our instance > > IDS2000 9.20.TC1 on Windows NT 4 Server SP 6a > > Have any of you seen this before ??? > > The database is using the TimeSeries datablade to capture financial info and > populating the data into a live database. We also have a test database on > the same instance (yes, yes I know...) that we are running a C program (MS > Visual C) to populate a table with simple data i.e. not TimeSeries, via > ODBC. > > We have loads of log space as described here:- > > Logical Logging > Buffer bufused bufsize numrecs numpages numwrits recs/pages pages/io > L-2 0 8 4776035 501490 458332 9.5 1.1 > Subsystem numrecs Log Space used > OLDRSAM 4776035 378148416 > > address number flags uniqid begin size used %used > c049eb8 1 U-B---- 294 700003 225000 32361 14.38 > c049ed4 2 U-B---- 295 c00003 25000 25000 100.00 > c049ef0 3 U-B---- 296 c061ab 25000 25000 100.00 > c049f0c 4 U-B---- 297 c0c353 25000 25000 100.00 > c049f28 5 U-B---- 298 c124fb 25000 17031 68.12 > c049f44 6 U---C-L 299 d00003 375000 368567 98.28 > c049f60 7 U-B---- 284 600035 1250 1250 100.00 > c049f7c 8 U-B---- 285 600517 1250 1250 100.00 > c049f98 9 U-B---- 286 6009f9 1250 487 38.96 > c049fb4 10 U-B---- 287 600edb 1250 1250 100.00 > c049fd0 11 U-B---- 288 6013bd 1250 1250 100.00 > c049fec 12 U-B---- 289 60189f 1250 1250 100.00 > c04a008 13 U-B---- 290 601d81 1250 1250 100.00 > c04a024 14 U-B---- 291 602263 1250 1250 100.00 > c04a040 15 U-B---- 292 602745 1250 1250 100.00 > c04a05c 16 U-B---- 293 602c27 1250 1250 100.00 > > > The problem we are getting is that the database appears to be performing a > long transaction roll back when hardly any of the log space is beng used - > at last count, the C program load used about 70,000 pages within log no. 6. > > We havn't modified the LTXHWM or LTXEHWM which are set to 50 & 60 > respectively > > Another funny, is that its not the C program that appears to be causing the > roll-back, its one of the VP's that Informix is running:- > > Informix Dynamic Server 2000 Version 9.20.TC1 -- On-Line -- Up 1 days > 00:00:48 -- 34240 Kbytes > > Message Log File: C:\\PROGRA~1\\informix\\ol_histinf.log > 16:57:15 Fuzzy Checkpoint Completed: duration was 0 seconds, 58 buffers> not flushed. > 16:57:15 Checkpoint loguniq 299, logpos 0x59f8c20c > > 16:57:15 Aborting Long Transaction: tx 0xc272e38 username: informix uid:6 > 16:58:14 Fuzzy Checkpoint Completed: duration was 0 seconds, 58 buffers> not flushed. > 16:58:14 Checkpoint loguniq 299, logpos 0x59f8e20c > > 16:58:14 Aborting Long Transaction: tx 0xc272e38 username: informix uid:6 > 17:00:10 Fuzzy Checkpoint Completed: duration was 0 seconds, 60 buffers> not flushed. > 17:00:10 Checkpoint loguniq 299, logpos 0x59f97aa0 > > 17:00:11 Aborting Long Transaction: tx 0xc273548 username: informix uid:6 > 17:00:14 Fuzzy Checkpoint Completed: duration was 0 seconds, 77 buffers> not flushed. > 17:00:14 Checkpoint loguniq 299, logpos 0x59f9c2a4 > > 17:00:15 Aborting Long Transaction: tx 0xc272728 username: informix uid:6 > 17:01:42 Fuzzy Checkpoint Completed: duration was 0 seconds, 58 buffers> not flushed. > 17:01:42 Checkpoint loguniq 299, logpos 0x59fad20c > > 17:01:42 Aborting Long Transaction: tx 0xc273548 username: informix uid:6 > > An onstat -u reveals:- > > > Userthreads > address flags sessid user tty wait tout locks nreads > nwrites > c244018 ---P--D 1 informix - 0 0 0 1365 > 1645 > c2445d4 ---P--F 0 informix - 0 0 0 0 > 174503 > c244b90 ---P--- 6 informix - 0 0 0 0 > 5349 > c24514c ---P--B 7 informix - 0 0 0 3703 > 6175 > c245cc4 ---P--D 11 informix - 0 0 0 4 1 > c246280 Y--P--- 13 informix HISTCL2 c44aa30 0 0 46756 10 > 6 active, 128 total, 14 maximum concurrent > > > > Sometimes, we are also getting this message from the onstat - command. When > we get this, we can't move the database onto the next logical log, or, in > fact, perform any onmode -l or -c commands. > > Informix Dynamic Server 2000 Version 9.20.TC1 -- On-Line(LONGTX)-- Up 1 > days 00:00:48 -- 34240 Kbytes > > > We have raised this with tech support, but havn't had any response back yet. > > All help much appreciated. > > Sean > > >