Re: [Condor-users] 6.7.20 windows condor_quill problem

[Horvátth Szabolcs <szabolcs@xxxxxxxxxxxxx>]

Hi Erik,

After having the problems for hours I had to turn the quill daemon off 
completely. I tried enabling it again today to
save some logs for you but it worked nicely for a few hours. Than its 
memory usage started growing
(from 4 mb went up to 10mb) and than "lost connection" with the scheduler.
Now if I query the state of the queue using condor_q I get an older 
state while condor_q -d schedd shows
the "real" queue. Its empty now btw so its quite easy to spot the 
difference.)

The only thing I found in the log file was this:

Now in new log file C:\Condor/log/QuillLog
7/3 17:47:51 Skipping over and going to the next job queue file
7/3 17:47:51 [QUILL] Unable to open the job_queue.log file!
7/3 17:47:51 Could not open file old job queue file 
C:\Condor/spool/job_queue.log.22989 --- ERROR
7/3 17:47:51 Skipping over and going to the next job queue file
7/3 17:47:51 [QUILL] Unable to open the job_queue.log file!
7/3 17:47:51 Could not open file old job queue file 
C:\Condor/spool/job_queue.log.22990 --- ERROR
7/3 17:47:51 Skipping over and going to the next job queue file
7/3 17:47:51 [QUILL] Unable to open the job_queue.log file!
7/3 17:47:51 Could not open file old job queue file 
C:\Condor/spool/job_queue.log.22991 --- ERROR
7/3 17:47:51 Skipping over and going to the next job queue file
7/3 17:47:51 [QUILL] ERROR: There is no such Proc Ad[22992.0]


The rest is just:

7/3 17:48:36 POLLING RESULT: NO CHANGE
7/3 17:48:36 ********* End of Probing Job Queue Log File *********
7/3 17:48:36 ++++++++ Sending schedd ad to collector ++++++++
7/3 17:48:36 ++++++++ Sent schedd ad to collector ++++++++
7/3 17:48:46 ******** Start of Probing Job Queue Log File ********
7/3 17:48:46 === Current Probing Information ===
7/3 17:48:46 fsize: 2697926        mtime: 1151941358
7/3 17:48:46 first log entry: 22992.0 WallClockCheckpoint 4711
7/3 17:48:46 POLLING RESULT: NO CHANGE
...

Cheers,
Szabolcs


Erik Paulson wrote:
> Could you put the full quill logfile on a webserver/ftp server and drop a
> note to condor-admin with the URL? 
>
> Thanks,
>
> -Erik
>