Active IQ Unified Manager Discussions

Controlling autosize threshold

kevin_graham
11,924 Views

Is there a way to control when FlexVol autosize is triggered? The documentation states that its when a volume is "about to run out of free space". However, once its crossed 95%, someone has already gotten woken up in the middle of the night and isn't about to trust that Magic Things Will Happen if free space falls any lower.

As-is, it seems to be a nice last-ditch effort, but I'd like to be able to set a threshold (~85-90%) for autosize to kick in to ensure that there's no false sense of security once utilization has started creeping way up and that its clear something really must be done to provide more space.

10 REPLIES 10

diarmuidn
11,880 Views

In priv set diag and you can reset the wafl_reclaim_threshold figure values using the 'setflag' command.

eg: setflag walf_reclaim_threshold_t = 96

(note: there are 5 values.. _t, _s, _m, _l, _xl. There represents tiny, small, medium, large and extra large volumes)

Use the printflag option to view current values

jimmcclung
11,880 Views

Can anyone provide what values constitute the various categories tiny, small, medium, large and extra large?

What I mean here is my 8 TB volume is an _xl volume. Into which category does an 100 GB volume fall? Or a 10 GB volume?

pascalduk
11,880 Views

I have the following list, but unfortunately I can't find the netapp source document anymore. It probably was in one of the tr's.

Volume size Trigger point
< 20 GB85%
20 - 100 GB90%
100 - 500 GB92%
500 GB - 1TB95%
> 1TB98%

jimmcclung
11,880 Views

Pascalduk nailed the answer to my question. Thanks!

radek_kubka
11,880 Views

Use the printflag option to view current values

To make it more fun, the output is actually in hex :


fas01*> printflag wafl_reclaim_threshold_t
wafl_reclaim_threshold_t = 0x55
fas01*> printflag wafl_reclaim_threshold_s
wafl_reclaim_threshold_s = 0x5a
fas01*> printflag wafl_reclaim_threshold_m
wafl_reclaim_threshold_m = 0x5c
fas01*> printflag wafl_reclaim_threshold_l
wafl_reclaim_threshold_l = 0x5f
fas01*> printflag wafl_reclaim_threshold_xl
wafl_reclaim_threshold_xl = 0x62

radek_kubka
11,880 Views

Also, changing thresholds via setflag is non-persistent & everything will change to the default values after reboot.

To make it persistent the applicable line(s) should be added to /etc/rc file on both controllers, e.g.:

priv set diag;setflag wafl_reclaim_threshold_xl 0x63;priv set

jimmcclung
11,880 Views

Thanks for the follow up!

henrypan1
11,877 Views

Radek,

I was wonder whether there is anything new in control autosize?

Thanks

Henry

radek_kubka
11,877 Views

Hi Henry,

I haven't come across anything new re autosize control - but maybe some NetApp insiders will chime in & tell us what's new or in the pipeline?

Arguably the method we know is a bit, hmm, clunky

regards,

Radek

wmulhair1
6,956 Views

Adding my voice to this.  Since our monitoring triggers alerts at 90% for any volume which results in somebody being paged in the middle of the night I'd rather have all volume autosizes triggers at 90% and respond to why the volume grew in the morning.

Public