Hi Max,
The issue here is that when the file was requested it wasn't found on
disk. So, the copy was queued. However, before actually tranfering the
file, the FRM checks if the file is already online. If so, the copy is
skipped and the client is told the file is available.
Andy
On Fri, 15 Dec 2017, Fischer, Max (SCC) wrote:
> Hi all,
>
> a new Tape buffer server recently started failing a lot of frm recalls because multiple `frm.xfr.copycmd in` were running for the same files. Judging from `frm_admin query xfrq`, from 6000 unique files queued in total, about 700 are queued more than once.
> Is there any known reason for such a behaviour? Can we clean up the queue somehow? Or must copycmd be able to handle the same request multiple times?
>
> OS: Scientific Linux release 7.3 (Nitrogen)
> xrootd: v4.7.1
>
> Cheers,
> Max
>
>
> ########################################################################
> Use REPLY-ALL to reply to list
>
> To unsubscribe from the XROOTD-L list, click the following link:
> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
>
########################################################################
Use REPLY-ALL to reply to list
To unsubscribe from the XROOTD-L list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
|