You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@trafficserver.apache.org by "John Plevyak (JIRA)" <ji...@apache.org> on 2010/02/10 01:43:27 UTC
[jira] Created: (TS-158) cache should not hold parititon lock over
callbacks/review locking
cache should not hold parititon lock over callbacks/review locking
------------------------------------------------------------------
Key: TS-158
URL: https://issues.apache.org/jira/browse/TS-158
Project: Traffic Server
Issue Type: Improvement
Components: Cache
Reporter: John Plevyak
Assignee: John Plevyak
Fix For: 2.1.0
Because we now have far fewer partitions it would be best
not to hold the partition lock over callbacks to reduce contention
on this lock and enable better scaling.
This is a good opportunity to review the locking in the cache as
well.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Resolved: (TS-158) cache should not hold parititon lock over
callbacks/review locking
Posted by "John Plevyak (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/TS-158?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
John Plevyak resolved TS-158.
-----------------------------
Resolution: Fixed
This was fixed as part of the cache work in -dev which is now in trunk.
> cache should not hold parititon lock over callbacks/review locking
> ------------------------------------------------------------------
>
> Key: TS-158
> URL: https://issues.apache.org/jira/browse/TS-158
> Project: Traffic Server
> Issue Type: Improvement
> Components: Cache
> Reporter: John Plevyak
> Assignee: John Plevyak
> Fix For: 2.1.0
>
>
> Because we now have far fewer partitions it would be best
> not to hold the partition lock over callbacks to reduce contention
> on this lock and enable better scaling.
> This is a good opportunity to review the locking in the cache as
> well.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (TS-158) cache should not hold parititon lock
over callbacks/review locking
Posted by "John Plevyak (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/TS-158?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12843724#action_12843724 ]
John Plevyak commented on TS-158:
---------------------------------
312 is
return (dir_tag(e) == DIR_MASK_TAG(key->word(2)));
both of which should be unsigned:
#define dir_tag(_e) ((_e)->w[2]&((1<<DIR_TAG_WIDTH)-1))
where inku16 w[5];
and MD5 word returns inku32
and
#define DIR_MASK_TAG(_t) ((_t) & ((1 << DIR_TAG_WIDTH) - 1))
???
> cache should not hold parititon lock over callbacks/review locking
> ------------------------------------------------------------------
>
> Key: TS-158
> URL: https://issues.apache.org/jira/browse/TS-158
> Project: Traffic Server
> Issue Type: Improvement
> Components: Cache
> Reporter: John Plevyak
> Assignee: John Plevyak
> Fix For: 2.1.0
>
>
> Because we now have far fewer partitions it would be best
> not to hold the partition lock over callbacks to reduce contention
> on this lock and enable better scaling.
> This is a good opportunity to review the locking in the cache as
> well.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (TS-158) cache should not hold parititon lock
over callbacks/review locking
Posted by "Leif Hedstrom (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/TS-158?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12843708#action_12843708 ]
Leif Hedstrom commented on TS-158:
----------------------------------
On RHEL4, we now get a compile error:
if ccache g++ -DHAVE_CONFIG_H -I. -I. -I../.. -I../../proxy -I../../librecords -I../../libinktomi++ -I../../iocore/eventsystem -I../../iocore/net -I../../iocore/aio -I../../iocore/hostdb -I../../iocore/cache -I../../iocore/cluster -I../../iocore/utils -I../../iocore/dns -I../../libev -I../../librecords -I../../proxy/http2 -I../../proxy/http2/remap -I../../proxy/hdrs -I../../proxy/mgmt2 -I../../proxy/mgmt2/preparse -I../../proxy/mgmt2/utils -I../../proxy/mgmt2/web2 -DSYSCONFDIR=\"/opt/ats/etc\" -I/usr/include -g -pipe -Wall -Werror -Wno-char-subscripts -Wno-write-strings -O3 -feliminate-unused-debug-symbols -fno-strict-aliasing -Wno-invalid-offsetof -march=i586 -MT Cache.o -MD -MP -MF ".deps/Cache.Tpo" -c -o Cache.o Cache.cc; \
then mv -f ".deps/Cache.Tpo" ".deps/Cache.Po"; else rm -f ".deps/Cache.Tpo"; exit 1; fi
In file included from P_Cache.h:41,
from Cache.cc:27:
P_CacheDir.h: In function `bool dir_compare_tag(Dir*, INK_MD5*)':
P_CacheDir.h:312: warning: comparison between signed and unsigned integer expressions
Cache.cc: In member function `int CacheVC::handleReadDone(int, Event*)':
Cache.cc:1798: warning: comparison between signed and unsigned integer expressions
Cache.cc: In member function `int CacheVC::handleRead(int, Event*)':
Cache.cc:1909: warning: comparison between signed and unsigned integer expressions
Cache.cc: In member function `int CacheVC::removeEvent(int, Event*)':
Cache.cc:2014: warning: comparison between signed and unsigned integer expressions
gmake[3]: *** [Cache.o] Error 1
gmake[3]: Leaving directory `/tmp/trafficserver.git/iocore/cache'
gmake[2]: *** [all-recursive] Error 1
gmake[2]: Leaving directory `/tmp/trafficserver.git/iocore'
gmake[1]: *** [all-recursive] Error 1
gmake[1]: Leaving directory `/tmp/trafficserver.git'
gmake: *** [all] Error 2
> cache should not hold parititon lock over callbacks/review locking
> ------------------------------------------------------------------
>
> Key: TS-158
> URL: https://issues.apache.org/jira/browse/TS-158
> Project: Traffic Server
> Issue Type: Improvement
> Components: Cache
> Reporter: John Plevyak
> Assignee: John Plevyak
> Fix For: 2.1.0
>
>
> Because we now have far fewer partitions it would be best
> not to hold the partition lock over callbacks to reduce contention
> on this lock and enable better scaling.
> This is a good opportunity to review the locking in the cache as
> well.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (TS-158) cache should not hold parititon lock over
callbacks/review locking
Posted by "Manjesh Nilange (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/TS-158?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Manjesh Nilange updated TS-158:
-------------------------------
Attachment: compiler-errors.patch
The typecasting that fixes these errors.
> cache should not hold parititon lock over callbacks/review locking
> ------------------------------------------------------------------
>
> Key: TS-158
> URL: https://issues.apache.org/jira/browse/TS-158
> Project: Traffic Server
> Issue Type: Improvement
> Components: Cache
> Reporter: John Plevyak
> Assignee: John Plevyak
> Fix For: 2.1.0
>
> Attachments: compiler-errors.patch
>
>
> Because we now have far fewer partitions it would be best
> not to hold the partition lock over callbacks to reduce contention
> on this lock and enable better scaling.
> This is a good opportunity to review the locking in the cache as
> well.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.