Grep'ing for a phone number in large Master.csv causes the sip peers to loose connectivity briefly

Asterisk Asterisk Support
1

folks,
i have a strange issue on asterisk 13.25.0 and centos 7:
i was grep ing for a phone number in /var/log/cdr-csv/Master.csv which was about 670MB; looks like it was not rotated in a while. this caused most of the sip peers to go unreachable and after a few seconds, they were reachable again.

the issue wont happen on smaller context.csv files

i was able to duplicate the issue after hours the load average as per top cmd was around 0.5 while i was running the grep cmd

i have since then rotated the Master.csv file

  1. anyone have any ideas?
  2. is this a known issue? should i be on the look out for any such large file searches?
  3. any suggestions, information is appreciated.

thanks

2

It sounds like you found the solution, manually, but to automate this you could consider a patch into CDR subsystem which emits warnings when there are significant delays in CDR processing, add a note to the cdr.conf.sample file about large CDR CSV files, or try out CDR batch mode to buffer CDRs (see cdr.conf for details.)

Non-Asterisk solutions are also available for watching file and directory sizes, emailing you alerts, etc.

3

Peers got unreachable by doing a grep search to a CSV file ?

4

Is this a virtual machine?

It sounds like there may be inadequate physical memory.

5

Thanks for the responses, yes, this is on a virtual machine.
According to fee and top cmds:
12137600kb total mem
305852kb free
1841100kb used
9990648kb buff/cached

6

You need the information from the host, not the virtual machine.

Ideally you wnat to set the virtual machine size to be a value that can be 100% committed against physical memory on the host. I think you are thrashing the host.

Also, I think the units are probably kB, not kb (factor of 8 difference)

1 Like
7

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.