[fossil-users] Is it possible to increase the sqlite timeout for fossil?
We are getting a *lot* of collisions in heavy use repos. Rather than confusing sync errors I'd like it if fossil just waited longer for the repo to come available. I think setting a 10 second timeout on sqlite would make a huge difference here. Can this be done? Thanks. ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Is it possible to increase the sqlite timeout for fossil?
On Thu, Feb 21, 2013 at 10:24 AM, Matt Welland estifo...@gmail.com wrote: We are getting a *lot* of collisions in heavy use repos. Rather than confusing sync errors I'd like it if fossil just waited longer for the repo to come available. I think setting a 10 second timeout on sqlite would make a huge difference here. Can this be done? Timeout is currently 5 seconds: http://www.fossil-scm.org/fossil/artifact/35a237cbfde?ln=723 You could change that and recompile. But before you do, first make sure you are in WAL mode. Can you post the /stat page for your repository - the equivalent of (http://www.fossil-scm.org/fossil/stat)? Thanks. ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users -- D. Richard Hipp d...@sqlite.org ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Is it possible to increase the sqlite timeout for fossil?
Interesting that the timeout is already 5 seconds. What happens is sync fails and then users are confused. Perhaps this is not a timeout? Note that this is direct file system access. I think suppressing the sqlite3 error would help and a try again? Y/n prompt. This is what we see: New_Version: 5df265af2129cc9d526860abc5897d084aa5ae25 Autosync: file://.blah.fossil Bytes Cards Artifacts Deltas Sent:5152 71 0 4 Error: Database error: unable to open database file INSERT INTO rcvfrom(uid, mtime, nonce, ipaddr)VALUES(1, julianday('now'), NULL, '127.0.0.1') Received: 159 1 0 0 Total network traffic: 2854 bytes sent, 348 bytes received fossil: Autosync failed Stat output: Repository Size: 214505472 bytes Number Of Artifacts: 18929 (stored as 7411 full text and 11518 delta blobs) Uncompressed Artifact Size: 108372 bytes average, 528150528 bytes max, 2051280245 bytes total Compression Ratio: 9:1 Number Of Check-ins: 3411 Number Of Files: 14198 Number Of Wiki Pages: 11 Number Of Tickets: 30 Duration Of Project: 743 days or approximately 2.03 years Project ID: b711128a80477cdb2f9a91b663dd6b2e8083c0b0 Server ID: b7aa6d92b218bd8086de653d1b7b58d7fac84c02 Fossil Version: 1.22 2012-03-19 12:45:47 [5dd5d39e7c] (gcc-4.4.1) SQLite Version: 2012-03-16 00:28:11 [74eadeec34] (3.7.11) Database Stats: 209478 pages, 1024 bytes/page, 113 free pages, UTF-8, delete mode On Thu, Feb 21, 2013 at 8:36 AM, Richard Hipp d...@sqlite.org wrote: On Thu, Feb 21, 2013 at 10:24 AM, Matt Welland estifo...@gmail.com wrote: We are getting a *lot* of collisions in heavy use repos. Rather than confusing sync errors I'd like it if fossil just waited longer for the repo to come available. I think setting a 10 second timeout on sqlite would make a huge difference here. Can this be done? Timeout is currently 5 seconds: http://www.fossil-scm.org/fossil/artifact/35a237cbfde?ln=723 You could change that and recompile. But before you do, first make sure you are in WAL mode. Can you post the /stat page for your repository - the equivalent of (http://www.fossil-scm.org/fossil/stat)? Thanks. ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users -- D. Richard Hipp d...@sqlite.org ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Is it possible to increase the sqlite timeout for fossil?
On Thu, Feb 21, 2013 at 11:21 AM, Matt Welland estifo...@gmail.com wrote: Database Stats: 209478 pages, 1024 bytes/page, 113 free pages, UTF-8, delete mode That delete mode notation on the end denotes the problem. You need to change your repository on the server to WAL mode. Do this as follows: sqlite3 $repository 'pragma journal_mode=wal' Or fossil rebuild $repository --pagesize 8192 --wal The second method will take (much) longer, but it also changes the page size from 1024 bytes to 8192 bytes, which won't make a huge difference but will help some. -- D. Richard Hipp d...@sqlite.org ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Is it possible to increase the sqlite timeout for fossil?
On Thu, Feb 21, 2013 at 9:25 AM, Richard Hipp d...@sqlite.org wrote: On Thu, Feb 21, 2013 at 11:21 AM, Matt Welland estifo...@gmail.com wrote: Database Stats: 209478 pages, 1024 bytes/page, 113 free pages, UTF-8, delete mode That delete mode notation on the end denotes the problem. You need to change your repository on the server to WAL mode. Do this as follows: sqlite3 $repository 'pragma journal_mode=wal' Or fossil rebuild $repository --pagesize 8192 --wal The second method will take (much) longer, but it also changes the page size from 1024 bytes to 8192 bytes, which won't make a huge difference but will help some. We can't switch to WAL mode currently as these repos are accessed from multiple machines via NFS. Now that ssh is working for us we can convert over to using ssh to a central server and switch to WAL mode at the same time. Is there any downside to using ssh? How dramatic is the performance improvement with WAL mode? Thanks. -- D. Richard Hipp d...@sqlite.org ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Is it possible to increase the sqlite timeout for fossil?
On Thu, Feb 21, 2013 at 11:45 AM, Matt Welland estifo...@gmail.com wrote: We can't switch to WAL mode currently as these repos are accessed from multiple machines via NFS. Now that ssh is working for us we can convert over to using ssh to a central server and switch to WAL mode at the same time. Is there any downside to using ssh? How dramatic is the performance improvement with WAL mode? The big advantage of WAL mode is that readers won't block writers. That can be a huge deal on a busy site. But the biggest gains you'll see will come from using ext4 instead of nfs as our filesystem. My gut instinct is that ssh will probably work better for this. Let me know if you encounter problems with it. -- D. Richard Hipp d...@sqlite.org ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users