Context Navigation

Timeline

12:41 Ticket #247 (obelix : error in main job header when using more than 1 node) created by jgipsl: Quand on dépasse 16 cpu MPI, l'entête du job est mal remplis. Par …
11:40 Changeset [1170] by aclsce: Modified to take into account modifications related to the use of sync …
11:40 Changeset [1169] by aclsce: Modified to take into account modifications related to the use of sync …

12:44 Changeset [1168] by sdipsl: Tag libIGCM_v2.5. This version brings several bug fixes and paves the road …
12:34 Milestone libIGCM_v2.5 metrics completed: Tag libIGCM_v2.5. This version brings several bug fixes and paves the road …
12:25 Ticket #246 (On Curie : rebuild on standard node using RAM high speed disk partition ; ...) closed by sdipsl: fixed: Done see r1167
12:22 Changeset [1167] by sdipsl: run rebuild on standard node using iRAM high speed disk partition
12:22 Changeset [1166] by sdipsl: the metrics template need more information create IGCM_comp_modifyFile …
12:20 Ticket #238 (Moving ins_job in libIGCM and finding a new/better name) closed by sdipsl: fixed: Done see r1164
12:19 Ticket #245 (On Curie machine we need to "/bin/sync" once per rebuild job) closed by sdipsl: fixed: Done see r1165
12:17 Ticket #246 (On Curie : rebuild on standard node using RAM high speed disk partition ; ...) created by sdipsl: Otherwise the rebuild is too slow. Too many single core process on the …

16:18 Changeset [1165] by sdipsl: Add a /bin/sync after the rebuild sequence (only active for curie). Should …
15:17 Changeset [1164] by sdipsl: Add ins_job in libIGCM. see #238 Dont change the name for now.

18:54 Changeset [1163] by sdipsl: - Add IGCM_sys_sync function. Do a /bin/sync on Curie and a dummy call …
18:53 Changeset [1162] by sdipsl: Send metrics results to a central place

14:17 Ticket #245 (On Curie machine we need to "/bin/sync" once per rebuild job) created by sdipsl: TGCC advises us to do so > De: "Yohan LEE-TIN-YIEN" <hotline.tgcc@…> …

17:58 Changeset [1161] by sdipsl: - this part is really ugly ... but let's continue to add to it for now …
17:54 Changeset [1160] by sdipsl: - better this way …
17:51 Changeset [1159] by sdipsl: - metrics first guess
17:48 Changeset [1158] by sdipsl: - metrics path
17:10 Changeset [1157] by sdipsl: - whitespace cleanup
16:50 Changeset [1156] by sdipsl: - homogenize indentation
16:45 Changeset [1155] by sdipsl: - enhanced comments - whitespace clean up
15:20 Ticket #241 (gencmip6 postprocessing must run on thin node) closed by sdipsl: fixed: Done see r1154. ins_job has been modified as well in the igcmg repo.

16:25 Ticket #243 (restarting from tar file broken at IDRIS) closed by sdipsl: fixed: Done see r1148
16:24 Ticket #244 (Avoid usage of tmpfile_$$) closed by sdipsl: fixed: Done see r1152
16:22 Changeset [1154] by sdipsl: - Add a pattern for node type. See #241
15:48 Changeset [1153] by sdipsl: - r1150 missed this bit. Need to flush messages when something wrong …
15:45 Changeset [1152] by sdipsl: - Avoid tmpfile usage. See #244

18:24 Ticket #244 (Avoid usage of tmpfile_$$) created by sdipsl: Deleted files remain visible for some time. chmod: cannot access …
17:40 Ticket #243 (restarting from tar file broken at IDRIS) created by sdipsl: mfls behaviour has been changed without notice. libIGCM can't find "tared" …
17:37 Ticket #242 (BigBrother on obelix) closed by sdipsl: fixed: Done see r1150
17:36 Ticket #237 (Enhance error handling affecting rebuild.ksh) closed by sdipsl: fixed: Done see r1151
17:34 Ticket #236 (AA_job Curie headers lacking openMP constraints) closed by sdipsl: fixed: Done see r1149

19:46 Changeset [1151] by sdipsl: adressing #237
19:20 Changeset [1150] by sdipsl: adressing #242 message 2000 and 3000 (pushstack and popstack) has been …
19:18 Changeset [1149] by sdipsl: adressing #236
16:56 Ticket #242 (BigBrother on obelix) created by sdipsl: /tmp is not a shared filesystem on obelix. Each obelix node have his own …
14:43 Ticket #241 (gencmip6 postprocessing must run on thin node) created by sdipsl: ins_job have to configure post processing jobs to run on thin node for the …

11:55 Ticket #240 (Explicit naming of rebuild jobs) closed by sdipsl: duplicate: this ticket and the #161 are the same

Note: See TracTimeline for information about the timeline view.