|
SA Bugzilla – Full Text Bug Listing |
Summary: | Spamc/Spamd very slow with -z compression and ssl | ||
---|---|---|---|
Product: | Spamassassin | Reporter: | Kevin A. McGrail <kmcgrail> |
Component: | spamc/spamd | Assignee: | SpamAssassin Developer Mailing List <dev> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | apache, giovanni, kmcgrail, mjcaley |
Priority: | P2 | ||
Version: | 3.4 SVN branch | ||
Target Milestone: | 4.0.0 | ||
Hardware: | All | ||
OS: | All | ||
Whiteboard: | |||
Attachments: |
Read only expected bytes
exit when data has been read |
Description
Kevin A. McGrail
2015-04-30 03:02:47 UTC
NOTE: Added a small debug for compress header acknowledgement to spamd so I could tell for sure that compression was working as expected. svn commit -m '3.4 commit - adding compress header debug statement - bug 7183' Sending spamd/spamd.raw Transmitting file data . Committed revision 1676888. svn commit -m 'trunk commit - adding compress header debug statement - bug 7183' Sending spamd/spamd.raw Transmitting file data . Committed revision 1676889. Continuing to target for 4.0 to look at this issue I've had some problems with doing this as well. It looks like the socket is reading a static number of bytes. Until the socket is closed it'll keep waiting for more data. I've made a patch that changes the read call to look for the expected number of bytes. Let me know what your thoughts are. Created attachment 5730 [details]
Read only expected bytes
Created attachment 5733 [details]
exit when data has been read
I think your diff will read all data at once which could be bad with big emails.
The attached diff will exit the sub as soon as enough data has been read.
Giovanni, do we need someone to test your revised patch? Yes, one more test could be useful' The problem was simply spamc itself, someone forgot to shutdown socket after sending data with SSL. Sending trunk/MANIFEST Sending trunk/spamc/libspamc.c Sending trunk/spamd/spamd.raw Adding trunk/t/spamd_ssl_z.t Transmitting file data ....done Committing transaction... Committed revision 1899775. |