snapmirror.readahead - what it does?

[aborzenkov]

I am currently facing issue of extremly slow snapmirror (1Gb/s direct line between two locations. ~2km apart). I noticed options:

snapmirror.readahead

snapmirror.readahead_freq

both with a value of 0 on a source filer. Anyone knows what these options do? They sound like somehing that could beef up snapmirror.

[radek_kubka]

The only bit of information I've found:

"Volume SnapMirror provides readahead tunables on the source filer to change the readahead on special WAFL files (active map files) that are needed to compute what blocks to send to the destination."

I would hazard a guess it won't help much with SnapMirror throughput - unless the source filer is constrained by some performance problems.

Regards,

Radek

[aborzenkov]

Thanks. I guess, it won’t help in this case – we have 12TB of data limping along with 25MB/s. Yes, source aggregate is busy. Attempt to enable VeryHigh priority for system IO did not help. May be at ~170 IOPS per data disk it won’t help much anyway. Unfortunately this has to finish till weekend and we do not have time to open case, because whatever happens it will never be resolved in time to finish (possibly, restarted) replication. We can only hope that source won’t run out of space until initialize is finished.

[scottgelb]

a BURT on it too depending on the source ontap version.  http://support.netapp.com/NOW/cgi-bin/bol?Type=Detail&Display=191374

[radek_kubka]

Yes, this is exactly where I took my quote from - the only search result on NOW site

[scottgelb]

Ahh...saw the copy/paste from there.  The search also brought up netapp ontap docs, but no listing of it...so indexed but no other info.  We need our own community site of hidden commands that a lot of us use but can't post