pigz / pkgacct high cpu
I'm seeing a very high CPU from pigz when the backup process is running:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
22280 root 39 19 11808 1268 560 R 94.4 0.0 346:57.55 pigz
15894 root 39 19 507m 413m 4352 S 17.9 1.3 50:12.94 pkgacct - allan
Its a large account, around 50Gb, so I'm not suprised that it will take a while to zip up, although the amount of time in the process is very concerning 50 hours for pkgacct and 348 hours for pigz...
What is concerning me most is the impact this is having on the web server. The load average is:
load average: 3.44, 2.39, 2.26
but even when its been down around 1, the load time has been exceptionally slow.
What can I do to improve this - both the zip time and more importantly keeping the server performant during the backup?
Thanks,
Allan
-
Hello Allan, The following options are available under the "Compression" tab in "WHM >> Tweak Settings": gzip compression level - The level of compression to use (0 is faster but less compression, 9 is slower but provides maximum compression) Number of pigz processes - The number of independent pigz processes to use when performing gzip compression. For quickest compression, specify the same number of cores available on your server. Number of kilobyte chunks per compression work unit - The size (in 1024 byte chunks) of compression work units to be distributed to each pigz process. Default is 4096. Systems with larger L2/L3 caches may benefit from higher values. Could you try modifying the values for these options for faster compression times (with less compression) to see if that helps during the next backup generation? Thank you. 0
Please sign in to leave a comment.
Comments
1 comment