Slow archive extract operations on Standard and Premium Filestore instances

[Ivan Stefanov]

Today we encounter an unexpected problem - during the deployment of our site we noticed that the .tgz archive extraction takes waaaay more time than expected when executed on Premium or Standard Filestore mounted on a Google Cloud Compute Engine VM. We've tested the extraction of a .tgz file with ~120MB size and results were quite disappointing: - Premium Filestore: 6m25.808s - Standard Filestore: 8m27.601s NB: Filestore and VM are located in the one and the same region and zones. NB: We do use the Recommended client machine type and recommended Linux client mount options at https://cloud.google.com/filestore/docs/performance For reference /proc/mounts looks as following: XX.XX.XX.XX:/shared /home/client/shared nfs rw,relatime,vers=3,rsize=524288,wsize=524288,namlen=255,hard,proto=tcp,timeo=600,retrans=2,sec=sys,mountaddr=XX.XX.XX.XX,mountvers=3,mountport=2050,mountproto=tcp,local_lock=none,addr=XX.XX.XX.XX 0 0 The expected result for this operation is to finish for less than 2 minutes. When comparing the extraction on another Network attached storage (not on GCP), on local standard disk or on SSD the results are: - NFS: 1m19.459s - local standard disk: 0m22.632s - SSD: 0m7.691s

Have you encountered such degraded performance by the Filestore and if yes is there any way to improve it?

[Ivan Stefanov]

[Alexis (Google Cloud Support)]

Hello Ivan,

I definitely agree that this should be faster. It doesn't look normal and the documentation also mentions that it should be faster, however there could be other factors affecting this.

I tried to reproduce your issue and I used a 60mb tar.gz file instead (half the size) and I did not have the same results. The extraction finished within a second. 6 or 8 minutes as you mentioned is not normal since that is slower than certain home hardware/connections and it's a small file. I suspect that this may be related to your project environment or the file itself somehow since I couldn't reproduce this (or maybe something else that I can't perceive at the moment). Typically, if I was able to recreate this, I would advise you to report the problem in the issue tracker here[1]. However, in this case it would be best to get in touch with one of our support reps if you have a package[2], as it would be more suitable for this kind of specific question.

If you do not have a package and you believe that I've missed something or that this is reproducible, the issue tracker[1] mentioned before still applies. The issue tracker also allows you to search for similar known issues. Hopefully this puts you in the right direction.

Sincerely,

Alexis

Google Cloud Support

[1] https://cloud.google.com/support/docs/issue-trackers

[2] https://cloud.google.com/support#tab2