Zdeněk Šustr [Mon, 30 Oct 2006 08:26:35 +0000 (08:26 +0000)]
db_actual_store() function introduced implementing identical parts of db_store() and db_parent_store() functions.
Miloš Mulač [Fri, 27 Oct 2006 12:41:46 +0000 (12:41 +0000)]
- load intJobStatus even when no events arrived yet (embrio subjob with lost registration event)
- 2 memory corruptions removed
Zdeněk Šustr [Fri, 27 Oct 2006 09:16:49 +0000 (09:16 +0000)]
Another improvement to Child Job Histrogram storage and retrieval functions.
Zdeněk Šustr [Fri, 27 Oct 2006 08:22:48 +0000 (08:22 +0000)]
Ammending a single error in edg_wll_GetSubjobHistogram()
Zdeněk Šustr [Thu, 26 Oct 2006 14:40:52 +0000 (14:40 +0000)]
JobID handling ammended in ChildHistogram storage and retrieval functions.
Miloš Mulač [Thu, 26 Oct 2006 11:52:19 +0000 (11:52 +0000)]
move collection stuff to jobstat (one level higher)
- mainly for sharing & debugging
Miloš Mulač [Thu, 26 Oct 2006 11:44:11 +0000 (11:44 +0000)]
separate function for storing CollectionState events
- needs some optimalizations
Miloš Mulač [Thu, 26 Oct 2006 11:35:43 +0000 (11:35 +0000)]
- switched to embrionic subjob registration
- some objects added for collection support
Miloš Mulač [Thu, 26 Oct 2006 11:30:00 +0000 (11:30 +0000)]
use string for child_event (easier for debugging)
Zdeněk Šustr [Tue, 24 Oct 2006 15:22:45 +0000 (15:22 +0000)]
Child histogram saving/retreival. Retreival not tested yet. Encoding/Decoding works all right.
Zdeněk Šustr [Tue, 24 Oct 2006 15:22:44 +0000 (15:22 +0000)]
Child histogram saving/retreival. Retreival not tested yet. Encoding/Decoding works all right.
Miloš Mulač [Mon, 23 Oct 2006 09:37:37 +0000 (09:37 +0000)]
increase version of int_status (now histograms encoded in)
Zdeněk Šustr [Fri, 20 Oct 2006 13:41:07 +0000 (13:41 +0000)]
Child job histogram encoding, storing and retrieval. Firt versions - retrieval not working yet.
Jan Pospíšil [Thu, 19 Oct 2006 22:31:15 +0000 (22:31 +0000)]
synchronized perftest outputs
Jan Pospíšil [Thu, 19 Oct 2006 17:09:50 +0000 (17:09 +0000)]
register subjobs directly
Miloš Mulač [Thu, 19 Oct 2006 14:35:24 +0000 (14:35 +0000)]
print correct JOBID indentier (useful for postprocessing scripts)
Miloš Mulač [Thu, 19 Oct 2006 14:30:01 +0000 (14:30 +0000)]
add NULL terminator at the end array (prevents misbehavior in producer's edg_wll_RegisterSubjobs)
Miloš Mulač [Thu, 19 Oct 2006 14:11:29 +0000 (14:11 +0000)]
count children states for parent job of collection too
Miloš Mulač [Thu, 19 Oct 2006 14:10:01 +0000 (14:10 +0000)]
print jobtype correctly
Miloš Mulač [Thu, 19 Oct 2006 13:19:30 +0000 (13:19 +0000)]
enable to register, process and print out job collections
Miloš Mulač [Tue, 17 Oct 2006 15:45:19 +0000 (15:45 +0000)]
added CollectionState event
- state and child_event encoded as ints (if strings preferable, let me know)
Miloš Mulač [Tue, 17 Oct 2006 11:52:16 +0000 (11:52 +0000)]
renamed GetSemID to more convenient JobSemphore
now returns true semphore number
Miloš Mulač [Tue, 17 Oct 2006 11:16:59 +0000 (11:16 +0000)]
don't lock parent if its jobid maps to same semaphore ID as children's one
- code is not active yet, so it is not tested
Miloš Mulač [Tue, 17 Oct 2006 07:55:17 +0000 (07:55 +0000)]
GetSemID added
- needs audit!
Miloš Mulač [Mon, 16 Oct 2006 09:37:36 +0000 (09:37 +0000)]
don't call update_parent_job yet (not finished, not working)
Jan Pospíšil [Sun, 15 Oct 2006 21:36:30 +0000 (21:36 +0000)]
Modified to compare several scenarios:
0) registration only to bkserver (edg_wll_RegisterJobSync)
1) dual registration (edg_wll_RegisterJobProxy)
2) old (not dual) registration (edg_wll_RegisterJobProxyOld)
3) two separate registrations (edg_wll_RegisterJobSync + edg_wll_RegisterJobProxyOnly)
Jan Pospíšil [Sun, 15 Oct 2006 21:16:42 +0000 (21:16 +0000)]
added edg_wll_RegisterJobProxyOld for testing/measuring (LB_PERF) purposes
Miloš Mulač [Thu, 12 Oct 2006 09:42:24 +0000 (09:42 +0000)]
after-merge cleanup
- seems working
Aleš Křenek [Wed, 11 Oct 2006 17:40:28 +0000 (17:40 +0000)]
default type plugin, including tests
Jan Pospíšil [Wed, 11 Oct 2006 17:25:03 +0000 (17:25 +0000)]
unit test to follow recent changes
Miloš Mulač [Wed, 11 Oct 2006 15:07:15 +0000 (15:07 +0000)]
small after-merge typos
Miloš Mulač [Wed, 11 Oct 2006 14:29:13 +0000 (14:29 +0000)]
better error/special situation handling in recv_token()
- prevent some rare-but-hard-to-catch errors
Aleš Křenek [Wed, 11 Oct 2006 13:35:47 +0000 (13:35 +0000)]
obsolete
Aleš Křenek [Tue, 10 Oct 2006 18:57:39 +0000 (18:57 +0000)]
builds after merge
Aleš Křenek [Tue, 10 Oct 2006 18:16:32 +0000 (18:16 +0000)]
- resolve duplicate -s
- adapt to new maildir interface in sandbox handling
Aleš Křenek [Tue, 10 Oct 2006 17:36:32 +0000 (17:36 +0000)]
merge from 3.1
František Dvořák [Fri, 6 Oct 2006 14:39:23 +0000 (14:39 +0000)]
Little change in DB module API - separate DB context creation from connection to DB.
Jan Pospíšil [Fri, 6 Oct 2006 12:24:51 +0000 (12:24 +0000)]
added -lcares to avoid some strange behaviour of the linker on some debian systems
Jan Pospíšil [Fri, 6 Oct 2006 12:18:47 +0000 (12:18 +0000)]
updated EDG_WLL_IL_PROTO errtext
Jan Pospíšil [Fri, 6 Oct 2006 12:17:00 +0000 (12:17 +0000)]
added DEBUG on demand
Jan Pospíšil [Fri, 6 Oct 2006 12:12:31 +0000 (12:12 +0000)]
hopefully last major update of the client library (at least for the moment)
- dualreg now should work (ifdefed under LB_PERF)
- added useful time measurements in the debug output
- huge clanup
Jan Pospíšil [Wed, 4 Oct 2006 06:11:21 +0000 (06:11 +0000)]
work in progress
- symbol names cleanup
- separate functions connect/write/read for all edg_wll_log{_proxy,_direct} versions
- some memleaks corrections
Jan Pospíšil [Sun, 1 Oct 2006 20:18:48 +0000 (20:18 +0000)]
huge code revision, rewritten almost everything
- separate functions
edg_wll_log_connect_proxy/direct
edg_wll_log_write_proxy/direct
edg_wll_log_read_proxy/direct
- megajob challenge - dual registration: RegisterJobProxy now tries to
do pseudo parallel registration to both bkserver and lbproxy
- compiles fine (LB_PERF), needs thorough testing
Miloš Mulač [Wed, 27 Sep 2006 14:20:44 +0000 (14:20 +0000)]
partial support for collections
- incomplete, mainly for sharing
Miloš Mulač [Wed, 27 Sep 2006 14:07:14 +0000 (14:07 +0000)]
add terminators (...NUMBER_OF...) for miscelaneous job status numeric codes enums
Miloš Mulač [Wed, 27 Sep 2006 14:04:09 +0000 (14:04 +0000)]
new jobtype COLLECTION
Miloš Mulač [Wed, 27 Sep 2006 14:02:28 +0000 (14:02 +0000)]
COLLECTION jobtype for RegisterEvent
Andrew McNab [Mon, 25 Sep 2006 13:13:59 +0000 (13:13 +0000)]
Sync
Andrew McNab [Mon, 25 Sep 2006 13:06:21 +0000 (13:06 +0000)]
Sync
Andrew McNab [Mon, 25 Sep 2006 12:25:38 +0000 (12:25 +0000)]
Sync
Andrew McNab [Mon, 25 Sep 2006 08:17:26 +0000 (08:17 +0000)]
Sync
Andrew McNab [Fri, 22 Sep 2006 13:41:08 +0000 (13:41 +0000)]
Try again
Andrew McNab [Fri, 22 Sep 2006 12:34:19 +0000 (12:34 +0000)]
Try again
Andrew McNab [Fri, 22 Sep 2006 09:56:29 +0000 (09:56 +0000)]
Try again
Andrew McNab [Fri, 22 Sep 2006 09:29:18 +0000 (09:29 +0000)]
Try again
Aleš Křenek [Fri, 22 Sep 2006 07:59:49 +0000 (07:59 +0000)]
push version
František Dvořák [Thu, 21 Sep 2006 17:56:35 +0000 (17:56 +0000)]
New jobid module megaupdate (replaced strmd5 by str2md5, work on ++ magic, ...).
Andrew McNab [Thu, 21 Sep 2006 16:52:39 +0000 (16:52 +0000)]
Try again
Andrew McNab [Thu, 21 Sep 2006 16:44:50 +0000 (16:44 +0000)]
Try again
Jan Pospíšil [Thu, 21 Sep 2006 09:53:05 +0000 (09:53 +0000)]
dualreg (work in progress)
Jan Pospíšil [Thu, 21 Sep 2006 09:27:35 +0000 (09:27 +0000)]
dateTime format update, microseconds should be separated by dot
Aleš Křenek [Thu, 21 Sep 2006 08:30:28 +0000 (08:30 +0000)]
on HEAD point to HEAD
Andrew McNab [Thu, 21 Sep 2006 08:12:37 +0000 (08:12 +0000)]
Try again
Jan Pospíšil [Thu, 21 Sep 2006 08:05:58 +0000 (08:05 +0000)]
aftermerge cleanup: duplicate VO, less FIXME
Jan Pospíšil [Thu, 21 Sep 2006 08:02:04 +0000 (08:02 +0000)]
ftermerge cleanup: removed mistakenly added already removed file
Andrew McNab [Wed, 20 Sep 2006 22:59:58 +0000 (22:59 +0000)]
Try again
Andrew McNab [Wed, 20 Sep 2006 21:37:06 +0000 (21:37 +0000)]
Try wtf for stdsoap2.h
Andrew McNab [Wed, 20 Sep 2006 20:20:04 +0000 (20:20 +0000)]
Fix grst_err krb5.h in Makefile
Andrew McNab [Wed, 20 Sep 2006 19:09:57 +0000 (19:09 +0000)]
Fix grst_err_globus in Makefile
František Dvořák [Wed, 20 Sep 2006 13:29:14 +0000 (13:29 +0000)]
Reset locale when depending on netstat output,
configurable IL socket/fileprefix via environment.
František Dvořák [Wed, 20 Sep 2006 12:26:13 +0000 (12:26 +0000)]
Changes forgotten in local CVS tree - it should compile now.
Zdeněk Salvet [Wed, 20 Sep 2006 10:50:08 +0000 (10:50 +0000)]
Remove CVS conflict section.
Jan Pospíšil [Wed, 20 Sep 2006 10:18:22 +0000 (10:18 +0000)]
less FIXMEs
Aleš Křenek [Tue, 19 Sep 2006 21:37:04 +0000 (21:37 +0000)]
Builds after merge _RC31
Aleš Křenek [Tue, 19 Sep 2006 20:57:04 +0000 (20:57 +0000)]
- favour GLITE_LOCATION
- install both perftests
Aleš Křenek [Tue, 19 Sep 2006 20:28:45 +0000 (20:28 +0000)]
Builds after merge _RC31
Daniel Kouřil [Tue, 19 Sep 2006 19:32:44 +0000 (19:32 +0000)]
transfered fix for #18880 from lb.common
if a message is split into several pieces during processing, the pieces aren't
unescaped independently but the whole message is reconstructed first before the
unescaping procedure is applied.
Aleš Křenek [Tue, 19 Sep 2006 18:14:44 +0000 (18:14 +0000)]
fix build
Daniel Kouřil [Tue, 19 Sep 2006 16:50:08 +0000 (16:50 +0000)]
Fix for #18880
if a message is split into several pieces during processing, the pieces aren't
unescaped independently but the whole message is reconstructed first before the
unescaping procedure is applied.
František Dvořák [Tue, 19 Sep 2006 14:02:33 +0000 (14:02 +0000)]
Transaction support to lb.proxy + migration script.
Reset locale in script depending on netstat output.
Michal Voců [Mon, 18 Sep 2006 15:41:43 +0000 (15:41 +0000)]
* made the test variants optional, see CVS log for perftest_il.sh for
description
Michal Voců [Mon, 18 Sep 2006 15:40:42 +0000 (15:40 +0000)]
* the test scripts can run selected tests only:
- options: -G <group> - selects group of tests
-t <test> - selects test variant
-n <num> - sets number of jobs
- <group> and <test> are single letters (run script without any options
to see their meaning)
- * means run all
- it is also possible to give space separated list
- example:
perftest_il -G a -t a
perftest_il -G \* -t "a x"
Michal Voců [Mon, 18 Sep 2006 15:35:40 +0000 (15:35 +0000)]
* sample job filenames are now stored in array, making the tests more
configurable (same filename in more fields repeats the same test,
there can be more than four sample jobs, if desired, etc.)
Andrew McNab [Tue, 12 Sep 2006 10:22:52 +0000 (10:22 +0000)]
sync
Andrew McNab [Thu, 7 Sep 2006 10:20:25 +0000 (10:20 +0000)]
Remove ap_compat: apr_table_unset/ap_table_unset
Andrew McNab [Tue, 5 Sep 2006 22:56:57 +0000 (22:56 +0000)]
Sync
Andrew McNab [Tue, 5 Sep 2006 22:53:19 +0000 (22:53 +0000)]
Sync
Jan Pospíšil [Mon, 4 Sep 2006 19:36:38 +0000 (19:36 +0000)]
updated README.LB-statistics
Jan Pospíšil [Mon, 4 Sep 2006 19:34:25 +0000 (19:34 +0000)]
*** empty log message ***
Jan Pospíšil [Sun, 3 Sep 2006 18:29:00 +0000 (18:29 +0000)]
do not duplicate records in lastStatusHistory
Jan Pospíšil [Sun, 3 Sep 2006 16:49:02 +0000 (16:49 +0000)]
added ident strings
Jan Pospíšil [Sun, 3 Sep 2006 15:52:58 +0000 (15:52 +0000)]
cosmetics
Jan Pospíšil [Sun, 3 Sep 2006 15:41:34 +0000 (15:41 +0000)]
cosmetics
Jan Pospíšil [Sun, 3 Sep 2006 12:50:41 +0000 (12:50 +0000)]
added JDL
Jan Pospíšil [Sun, 3 Sep 2006 09:45:08 +0000 (09:45 +0000)]
- memory allocation cleanup
- merged changes from glite-lb-server_branch_1_4_0_RC31
Jan Pospíšil [Sun, 3 Sep 2006 09:41:48 +0000 (09:41 +0000)]
added (previosly somehow lost) attribute VO
Jan Pospíšil [Sun, 3 Sep 2006 09:24:13 +0000 (09:24 +0000)]
added support of JDL attribute (only by an extra -j option)
Jan Pospíšil [Sat, 2 Sep 2006 21:58:54 +0000 (21:58 +0000)]
updated fullStatusHistory and lastStatusHistory
- lastStatusHistory now subset of fullStatusHistory (the first item is the same - Submitted,
but the rest is only from the last Waiting state to the end)
- reasons now properly filled in
- timestamps still have the usec equal to zero, the only way to change it is probably
to modify the type of timestamp in glite_jp_attrval_t from time_t to struc timeval?
František Dvořák [Fri, 1 Sep 2006 14:26:23 +0000 (14:26 +0000)]
Added the purge dir yet, the dump dir needs to be cleaned.
František Dvořák [Fri, 1 Sep 2006 14:17:32 +0000 (14:17 +0000)]
Merge forgotten changes from other branch, mark other two patches obsolete.