PICARD 60b

Summary

- Taking a 5s nap before copying and opening the jobReport.json file.

Problem seen with some jobs, e.g. 2397783893 where the jobReport.json file

was cut off, indicating that it was not finished writing to

- Fixed curl handling in getFAXRedirectors(). Previously 'None' was not

handled correctly leading to the occasional bad curl call. Requested by

Ilija Vukotic

- Added new AthenaMP error code for event service: ERR_ESATHENAMPDIED =

1227, "AthenaMP ended Event Service job prematurely"

- In debug mode, now uploading the latest *.log file (if present) instead

of stdout. Requested by Rod Walker

Update info from Edward Karavakis:

- Added error handling for potential glExec failures

- Added a retry mechanism with a sleep and if glexec fails, don't break

the pilot

- Update panda server when glexec infrastructure fails

- Added a new pilot error code for glexec related failures: 1226, ³glExec

related error"

- List dirs and copy them instead of having them hardcoded

- Corrected glexec_utils 2>&1 instead of 2>1

- Do not copy pilot proxy to /tmp when pinging glexec infrastructure, use

X509_USER_PROXY instead

Update info from Wen:

- Added new site mover for S3 and added it to the site mover farm (this

means that the S3 object store can be used as a ³normal² site mover)

- Updates to S3ObjectstoreSiteMover related to checksum handling (only md5

can be used with S3 object store, but ad32 is calculated as well)

- Added missing setting of stdout in run(), fixing the problem with the

time-out function 

Update info from Vincenzo Lavorini:

- Some error handling added to aria2c site mover

- Added protections around some of the file open commands in the aria2c

site mover

Version info

Taking a 5s nap before copying and opening the jobReport.json file to make sure it is finished, in getTrfExitInfo(). Problem seen with some jobs, e.g. 2397783893

where the jobReport.json file was cut off, indicating that it was not finished writing to (RunJob)

Added radical and HPC to directories to be copied to sandbox, in __ship_queue_data() (glexec_utils)

Fixed curl handling in getFAXRedirectors(). Previously 'None' was not handled correctly (FAXSiteMover)

Now sending jobId to findGlobalFilePath() from get_data() (FAXSiteMover)

Added optional jobId argument to findGlobalFilePath(), getGlobalFilePaths() (FAXSiteMover)

Sending jobId to getGlobalFilePaths() from findGlobalFilePath() (FAXSiteMover)

Sending jobId to getFAXRedirectors() from getGlobalFilePaths() (FAXSiteMover)

Added jobId argument to getFAXRedirectors() (FAXSiteMover)

Added new module that will contain functions related to file handling (FileHandling)

getStdoutFilename() now using function from FileHandling to find last updated log file to be used in debug mode (pUtil)

Added debug info to createLogFile() (JobLog)

Now setting pilotErrorDiag if there is no path to the object store in getDDMStorage() (Mover)

Added new error code: ERR_ESATHENAMPDIED = 1227, "AthenaMP ended Event Service job prematurely" (RunJobEvent)

Added error handling of event service failures in __main__() (RunJobEvent)

Stopping threads after various failures in __main__() (RunJobEvent)

Using pandaserver instead of dev server aipanda007 in updateEventRange() and downloadEventRanges() (EventRanges)

Using pandaserver instead of dev server aipanda007 in downloadEventRanges() (RunJobEvent)

Update info from Edward Karavakis:

- Added a retry mechanism with a sleep and if glexec fails, don't break the pilot

- Update panda server when glexec infrastructure fails

- Added a new pilot error code for glexec related failures: 1226

- List dirs and copy them instead of having them hardcoded (that's why glexec broke with HPC, radical)

- Corrected glexec_utils 2>&1 instead of 2>1

- Do not copy pilot proxy to /tmp when pinging glexec infrastructure, use X509_USER_PROXY instead

Update info from Wen:

- getCopyPrefixPath() updated for ..

- Added new site mover for S3 and added it to the site mover farm (S3SiteMover, SiteMoverFarm)

- Updates to S3ObjectstoreSiteMover related to checksum handling

- Added missing setting of stdout in run() (TimerCommand)

Update info from Vincenzo Lavorini:

- Some error handling added to aria2c site mover

- Added some protections around file open (aria2cSiteMover)