LISTSERV mailing list manager LISTSERV 16.5

Help for XROOTD-L Archives


XROOTD-L Archives

XROOTD-L Archives


XROOTD-L@LISTSERV.SLAC.STANFORD.EDU


View:

Message:

[

First

|

Previous

|

Next

|

Last

]

By Topic:

[

First

|

Previous

|

Next

|

Last

]

By Author:

[

First

|

Previous

|

Next

|

Last

]

Font:

Proportional Font

LISTSERV Archives

LISTSERV Archives

XROOTD-L Home

XROOTD-L Home

XROOTD-L  September 2019

XROOTD-L September 2019

Subject:

Re: Help debugging slow Xrootd Proxy Cache <-> xrootd server transfers

From:

Matevz Tadel <[log in to unmask]>

Reply-To:

Support use of xrootd by HEP experiments <[log in to unmask]>

Date:

Fri, 6 Sep 2019 15:32:15 -0700

Content-Type:

text/plain

Parts/Attachments:

Parts/Attachments

text/plain (289 lines)

Sorry, haven't noticed this before ... but I have noticed the issue report and 
fixed it :)

\m

On 9/6/19 12:10 PM, Andrew Hanushevsky wrote:
> I don't know the reason why there is a 16M limit and it clearly is not ideal for 
> Ceph or HDFS type storage systems. Matevz, do you know? Can the limit be removed?
> 
> Andy
> 
> On Fri, 6 Sep 2019, Sam Skipsey wrote:
> 
>> Hm. Adding that line to the config prevents the xrootd cache service from
>> staying up (and removing it makes it work again).
>>
>> 190906 19:03:37 19779 XrdFileCache_a2x: get block size 64M may not be
>> greater than 16777216
>> 190906 19:03:37 19779 XrdFileCache_Manager: error Cache::Config() error in
>> parsing
>>
>> Setting it to the maximum,
>>
>> pfc.blocksize 16M
>>
>> does significantly improve the rate though (I get about 80 to 90% of the
>> direct rate - around 300MB/s).
>>
>> thanks.
>>
>> Is it possible to have the arbitrary limit on pfc.blocksize removed in
>> later releases?
>>
>> Sam
>>
>>
>>
>> On Fri, 6 Sep 2019 at 18:03, Yang, Wei <
>> [log in to unmask]> wrote:
>>
>>> Hi Sam,
>>>
>>>
>>>
>>> One thing that maybe useful for Xcache to talk to a CEPH storage is to set
>>> the Xcache "*pfc.blocksize*" to the same as the CEPH native bucket size.
>>> So if the CEPH bucket size is 64M, then you can add the following to
>>> cephc02's config
>>>
>>>
>>>
>>> pfc.blocksize 64M
>>>
>>>
>>>
>>> The default is 1M
>>>
>>>
>>>
>>> regards,
>>>
>>> -- 
>>>
>>> Wei Yang  |  [log in to unmask]  |  650-926-3338 (O)
>>>
>>>
>>>
>>>
>>>
>>> *From: *<[log in to unmask]> on behalf of Sam Skipsey <
>>> [log in to unmask]>
>>> *Date: *Friday, September 6, 2019 at 6:20 AM
>>> *To: *xrootd-l <[log in to unmask]>
>>> *Subject: *Help debugging slow Xrootd Proxy Cache <-> xrootd server
>>> transfers
>>>
>>>
>>>
>>> Hello everyone:
>>>
>>>
>>>
>>> I'm currently seeing some odd behaviour, and would appreciate some insight
>>> from people more deeply aware of xrootd than I am.
>>>
>>>
>>>
>>> Currently we are testing an xrootd setup using internal Xrootd disk
>>> caching proxies to improve performance against a Ceph object store.
>>>
>>>
>>>
>>> The configuration is:
>>>
>>>
>>>
>>> Xrootd server [with ceph plugin] on cephs02.beowulf.cluster
>>>
>>>
>>>
>>> Xrootd cache on cephc02.beowulf.cluster
>>>
>>> Cache is backed by a software raid-0 array of 6 SSDs. The SSD array
>>> achieves > 1.5GB/s transfer rates when tested, and is not an I/O bottleneck.
>>>
>>> Cephc02 is configured as a direct proxy for cephs02.
>>>
>>>
>>>
>>> Firewall is configured so all nodes are in the trusted zone relative to
>>> each other, so no ports blocked.
>>>
>>>
>>>
>>> The problem is that connections proxied through cephc02's xrootd server
>>> are extremely slow (10x to 15x slower) than direct connections to the
>>> xrootd server on cephs02.
>>>
>>>
>>>
>>> From cephc02, directly copying from cephs02:
>>>
>>>
>>>
>>> [root@cephc02 ~]# time xrdcp root://cephs02:1095/ecpool:testfile2GB
>>> testfile2GB-in-1
>>>
>>> [1.863GB/1.863GB][100%][==================================================][381.5MB/s] 
>>>
>>>
>>>
>>>
>>> versus connecting via the proxy on cephc02 (cold cache):
>>>
>>> [root@cephc02 ~]# time xrdcp -v -v -v root://
>>> 10.1.50.12:1094/ecpool:testfile2GB test-from-cache2
>>>
>>> [1.863GB/1.863GB][100%][==================================================][20.51MB/s] 
>>>
>>>
>>>
>>>
>>> (once the cache is warm, fetching from the cache itself is v fast, at >
>>> 1GB/s)
>>>
>>>
>>>
>>>
>>>
>>> Whilst I'd expect some caching overhead, this seems unuseably slow.
>>>
>>> What am I doing wrong here?
>>>
>>>
>>>
>>> Any help appreciated,
>>>
>>>
>>>
>>> Sam Skipsey
>>>
>>> University of Glasgow
>>>
>>>
>>>
>>>
>>>
>>> The Cache and Server authenticate with a shared secret, and their relevant
>>> configs are:
>>>
>>>
>>>
>>> cephc02:
>>>
>>>
>>>
>>> [root@cephc02 ~]# cat /etc/xrootd/xrootd-cache.cfg
>>> ofs.osslib    libXrdPss.so
>>> pss.cachelib  libXrdFileCache.so
>>> pfc.ram      16g
>>> pfc.trace     info
>>> pfc.diskusage 0.90 0.95
>>> oss.localroot /cache/
>>>
>>> all.export /xroot:/
>>> all.export /root:/
>>> all.export *
>>> pss.origin 10.1.50.2:1095
>>> xrd.allow host *.beowulf.cluster
>>>
>>> #inbound security protocol, for authenticating to the xrootd-ceph
>>> #sec.protocol
>>> xrootd.seclib /usr/lib64/libXrdSec.so
>>> sec.protocol sss -s /etc/gateway/xrootd/sss.keytab.grp -c
>>> /etc/gateway/xrootd/sss.keytab.grp
>>> sec.protbind cephs02.beowulf.cluster:1095 only sss
>>>
>>> #outside security protocol, for authenticating users wanting to use the
>>> proxy
>>> #sec.protocol
>>> sec.protbind localhost only none
>>>
>>> xrd.report 127.0.0.1:9527 every 5s all
>>>
>>>
>>>
>>> -
>>>
>>>
>>>
>>> cephs02:
>>>
>>>
>>>
>>> # The export directive indicates which paths are to be exported. While the
>>> # default is '/tmp', we indicate it anyway to show you this directive.
>>> #
>>> all.export *?
>>> all.export /
>>>
>>> # The adminpath and pidpath variables indicate where the pid and various
>>> # IPC files should be placed
>>> #
>>> all.adminpath /var/spool/xrootd
>>> all.pidpath /var/run/xrootd
>>> xrootd.async segsize 67108864
>>> xrd.buffers maxbsz 67108864
>>>
>>> # Configure sss security - this is a shared secret between the xrootd-ceph
>>> and the xrootd-proxies, so the proxies are trusted to talk to the
>>> xrootd-ceph
>>> #
>>> xrootd.seclib /opt/xrootd/lib64/libXrdSec.so
>>> sec.protocol sss -s /etc/gateway/xrootd/sss.keytab.grp -c
>>> /etc/gateway/xrootd/sss.keytab.grp
>>> sec.protbind * only sss
>>>
>>> xrootd.seclib /opt/xrootd/lib64/libXrdSec.so
>>> #sec.protocol host
>>> #sec.protbind localhost none
>>>
>>> # Configure rados connection
>>> #   /this needs to be configured for the right stripe width
>>> ofs.osslib +cksio /opt/xrootd/lib64/libXrdCeph.so admin@ecpool
>>> ,1,8388608,83886080
>>> ofs.xattrlib /opt/xrootd/lib64/libXrdCephXattr.so
>>> xrootd.chksum adler32
>>>
>>> # Configure the port
>>> #
>>> xrd.port 1095
>>>
>>>
>>>
>>>
>>> ------------------------------
>>>
>>> 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
>>>
>>
>> ########################################################################
>> 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

########################################################################
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

Top of Message | Previous Page | Permalink

Advanced Options


Options

Log In

Log In

Get Password

Get Password


Search Archives

Search Archives


Subscribe or Unsubscribe

Subscribe or Unsubscribe


Archives

June 2024
May 2024
April 2024
March 2024
February 2024
January 2024
December 2023
November 2023
October 2023
September 2023
August 2023
July 2023
June 2023
May 2023
April 2023
March 2023
February 2023
January 2023
December 2022
October 2022
September 2022
August 2022
July 2022
June 2022
May 2022
April 2022
March 2022
February 2022
January 2022
December 2021
November 2021
October 2021
September 2021
August 2021
July 2021
June 2021
May 2021
April 2021
March 2021
February 2021
January 2021
December 2020
November 2020
October 2020
September 2020
August 2020
July 2020
June 2020
May 2020
April 2020
March 2020
February 2020
January 2020
December 2019
November 2019
October 2019
September 2019
August 2019
July 2019
June 2019
May 2019
April 2019
March 2019
February 2019
January 2019
December 2018
November 2018
October 2018
September 2018
August 2018
July 2018
June 2018
May 2018
April 2018
March 2018
February 2018
January 2018
December 2017
November 2017
October 2017
September 2017
August 2017
July 2017
June 2017
May 2017
April 2017
March 2017
February 2017
January 2017
December 2016
November 2016
October 2016
September 2016
August 2016
July 2016
May 2016
April 2016
March 2016
February 2016
January 2016
December 2015
November 2015
October 2015
September 2015
August 2015
July 2015
June 2015
May 2015
April 2015
March 2015
February 2015
January 2015
December 2014
November 2014
October 2014
September 2014
August 2014
July 2014
June 2014
May 2014
April 2014
March 2014
February 2014
January 2014
December 2013
November 2013
October 2013
September 2013
August 2013
July 2013
June 2013
May 2013
April 2013
March 2013
February 2013
January 2013
December 2012
November 2012
October 2012
September 2012
August 2012
July 2012
June 2012
May 2012
April 2012
March 2012
February 2012
January 2012
December 2011
November 2011
October 2011
September 2011
August 2011
July 2011
June 2011
May 2011
April 2011
March 2011
February 2011
January 2011
December 2010
October 2010
September 2010
August 2010
July 2010
June 2010
May 2010
April 2010
March 2010
February 2010
January 2010
December 2009
November 2009
October 2009
September 2009
July 2009
June 2009
May 2009
April 2009
March 2009
January 2009
December 2008
September 2008
August 2008
July 2008
June 2008
May 2008
April 2008
March 2008
February 2008
January 2008
December 2007
November 2007
October 2007
September 2007
August 2007
July 2007
June 2007
May 2007
April 2007
March 2007
February 2007
January 2007
December 2006
November 2006
October 2006
September 2006
August 2006
July 2006
June 2006
May 2006
April 2006
March 2006
February 2006
January 2006
December 2005
November 2005
October 2005
September 2005
August 2005
July 2005
June 2005
May 2005
April 2005
March 2005
February 2005
January 2005
December 2004
November 2004
October 2004
September 2004
August 2004

ATOM RSS1 RSS2



LISTSERV.SLAC.STANFORD.EDU

Secured by F-Secure Anti-Virus CataList Email List Search Powered by the LISTSERV Email List Manager

Privacy Notice, Security Notice and Terms of Use