Print

Print


Hi James,

I would strongly advise against turning off pgread/pgwrite as then we 
would not be able to detect data corruption. I don't think that would be 
wise. As for 64KB chunks, our tests have shown that over the WAN this is 
an optimal size to allow maximum I/O overlap and achieve sustained 
streaming. BTW be aware that OpenSSL also uses a 64KB block size and no 
one has complained about that. But if that's what you want see the
pss.cschk directive at

https://xrootd.slac.stanford.edu/doc/dev53/pss_config.htm#_Toc75537973

and

https://xrootd.slac.stanford.edu/doc/dev53/pss_config.htm#_Toc75537975

Andy

On Fri, 17 Sep 2021, James William Walder wrote:

> Hi,
> Is I possible (and how) to disable pgReads and Writes in an Xrootd server / Xcache config?
>
> I?m not sure if the setting can be done on a Mem/Xcache side, or on the server, or perhaps both?
>
> I?m running some tests with 5.3.1 with a proxy + Server( where the server is using the XrdCeph plugin), and am seeing the pgReads appearing.
> However, I?m not sure if 64kb reads is particularly optimal in this case, and it would be nice to have some options to disable it if possible.
>
> Thanks in advance,
> James
>
>
>
>
> ########################################################################
> 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