Re: Diag dest

I get your point about NFS, but, mainly, in RAC systems you use ASM (cellnodes, san/nas, whatever) and your DB files reside there. If it "freezes" then you will have a bigger problem that diag dest most likely.
Also having these files in an ACFS allows you to access all files from any single node, instead of having to jump between nodes. Isn't that something to consider too?

On Wed, Jul 12, 2023 at 11:18 AM Mark W. Farnham <mwf_at_rsiz.com> wrote:

> Nods. The last thing you want to do is impair performance because of
> diagnostics getting hung up on network issues. +42 to Ilmar.
>
>
>
> in re: inode limits. It may be a pain in the butt, but I recommend an at
> least daily sweeper: make a subdirectory yyyymmdd in each accumulating log
> and diagnostic directory and on yyyymmdd+2 mv all yyyymmdd files into
> yyyymmdd.
>
>
>
> the mv local to a subdirectory moves only the names, but each subdirectory
> gets its own inode limit and you’re not going to run out of days. Then if
> you want to centralize with an actual copy of the contents you have a
> useful container to move. Organization by time (date) has always been
> useful. Purging is also efficient.
>
>
>
> Some cases may require more frequent sweeping, but daily after a waiting a
> full day is a useful default configuration.
>
>
>
> mwf
>
>
>
> *From:* oracle-l-bounce_at_freelists.org [mailto:
> oracle-l-bounce_at_freelists.org] *On Behalf Of *Ilmar Kerm
> *Sent:* Tuesday, July 11, 2023 4:35 PM
> *To:* ricard.martinez_at_gmail.com
> *Cc:* Oracle-L Freelists
> *Subject:* Re: Diag dest
>
>
>
> Local only local.
>
> ACFS, NFS (anything shared?) can freeze and there also goes your database
> instance.
>
> Also have hit some inode limits for NAS systems, that Oracle diag dest (or
> old style audit dest) will eventually reach.
>
>
>
> Have a local agent on each server that will send the file contents to a
> central destination (like splunk or just a time series database)
>
>
>
> Ilmar
>
>
>
> On Tue, Jul 11, 2023 at 9:51 PM Ricard Martinez <ricard.martinez_at_gmail.com>
> wrote:
>
> Hi everyone,
>
>
>
> I've worked in several organizations with radical different visions about
> where to store the audit and trace files.
>
> RAC nodes using ACFS for diag to have access to all files from any nodes,
> others using local filesystem for the same.
>
> What are you thought and/or cons/pros about it?
>
> Do you use ACFS as diag destination? Do you prefer to use local filesystem?
>
>
>
> Thanks
>
>
>
>
> --
>
> Ilmar Kerm
>

--
http://www.freelists.org/webpage/oracle-l