Date: Wed, 13 May 2009 16:14:40 +1000 From: David N <davidn04@gmail.com> To: questions@freebsd.org Subject: Re: GJournal: Out of cache force switch - With lock up Message-ID: <4d7dd86f0905122314g671f3f91v1c1d646a24fab77f@mail.gmail.com> In-Reply-To: <4d7dd86f0905122254u5aa0f7aas32bb72dd52e3919@mail.gmail.com> References: <4d7dd86f0905122254u5aa0f7aas32bb72dd52e3919@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
2009/5/13 David N <davidn04@gmail.com>: > Hi, > > I'm running geom journal in debug and I'm seeing some > > GEOM_JOURNAL[1]: No cache, waking up the switcher. > GEOM_JOURNAL[1]: Out of cache, force switch (used=...... limit=...... ) > > But on the journal i left 2GB, how come its running out of cache? or > is that the journal cache that it flushes to disk? > > > I've also managed to lock up the system, and just trying to pin point > the problem. This is a 7.1R just upgraded to 7.2 with GPT + gmirror + > gjournal with GENERIC kernel. > All SSH no data. Local console is locked as well, but the num lock > light turns on/off. > > I had gstat running, and all disk activity went to zero before it locked. > > I'm just cp -av from one gmirror to the new gmirror with journal. > > I've done it twice so far, with about 1 hour of copying and it'll lock up. > > > Regards > David N > It happened again GEOM_JOURNAL[1]: Out of cache, force switch (used=175878656 limit=167772160). GEOM_JOURNAL[1]: Msync time of /usr: 0.156931s GEOM_JOURNAL[1]: Sync time of /usr: 0.006427s GEOM_JOURNAL[1]: Cache flush time: 0.267372s GEOM_JOURNAL[1]: BIO_FLUSH time of mirror/gm0p6: 0.267709s GEOM_JOURNAL[1]: Data has been copied. GEOM_JOURNAL[1]: Data has been copied. GEOM_JOURNAL[1]: Suspend time of /usr: 31.412925s GEOM_JOURNAL[1]: Starting copy of journal. GEOM_JOURNAL[1]: Cache flush time: 0.290278s GEOM_JOURNAL[1]: Cache flush time: 0.000101s GEOM_JOURNAL[1]: Switch time of mirror/gm0p6: 0.296598s GEOM_JOURNAL[1]: Msync time of /var: 0.000031s GEOM_JOURNAL[1]: Sync time of /var: 0.000048s GEOM_JOURNAL[1]: Cache flush time: 0.395199s GEOM_JOURNAL[1]: BIO_FLUSH time of mirror/gm0p5: 0.395694s GEOM_JOURNAL[1]: Suspend time of /var: 0.000226s GEOM_JOURNAL[1]: Starting copy of journal. GEOM_JOURNAL[1]: Cache flush time: 0.708414s GEOM_JOURNAL[1]: Cache flush time: 0.912773s GEOM_JOURNAL[1]: Switch time of mirror/gm0p5: 3.372491s GEOM_JOURNAL[1]: Entire switch time: 35.917632s managed to copy this before it locked up, the used is greater than the limit... Is there anything in sysctl i can tune to make it switch quicker? The last debug message is Data has been copied. Regards David N
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4d7dd86f0905122314g671f3f91v1c1d646a24fab77f>