From mboxrd@z Thu Jan 1 00:00:00 1970 From: Edward Shishkin Subject: Re: reiser4 mount options Date: Mon, 24 Dec 2007 17:28:39 +0300 Message-ID: <476FC217.9090408@namesys.com> References: <14476466.post@talk.nabble.com> <476E6CE2.7030004@namesys.com> <14483298.post@talk.nabble.com> <476EFF85.2070002@namesys.com> <14484849.post@talk.nabble.com> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <14484849.post@talk.nabble.com> Sender: reiserfs-devel-owner@vger.kernel.org List-ID: Content-Type: text/plain; charset="us-ascii"; format="flowed" To: butdiene Cc: reiserfs-devel@vger.kernel.org butdiene wrote: >I think this is a success report. > >with optimal_io_size=2097152,atom_max_size=0x800, reiser4 runs very >reasonable now. I put very heavy load on the file system. In past, those >operations would definitely hang firefox for minutes, but it's barely >noticeable now. > >fstab line: >UUID=1b9fdd32-127f-4751-b6aa-5eb300000000 / reiser4 >defaults,iocharset=utf8,optimal_io_size=2097152,atom_max_size=0x800 0 1 > >cat /proc/mounts: >/dev/root / reiser4 >rw,atom_max_size=0x2f238,atom_max_age=0x927c0,atom_min_size=0x100,atom_max_flushers=0x1,cbk_cache_slots=0x10 >0 0 > >/proc/mounts doesn't agree with either fstab or "mount", I don't know why. I >dare not to modify reiser4 source to print out valid settings. > > > Valid settings are in /proc/mounts, so we can see, that the problem still exists.. Sorry, ignore my suggestion about fstab. It is impossible to pass reiser4-specific mount options for ''/" via fstab, because all such options are passed to "/" via remount, but ->remount_fs() super operation is not implemented for reiser4. However, it is possible to pass reiser4-specific mount options for "/" via boot loader. I have passed a string "rootflags=tmgr.atom_max_size=2048" to my grub, and it works for me. Note, that prefix "tmgr." is required and values should be passed in decimal format. Thanks, Edward. >thanks a lot! > > >output of mount: >/dev/sda3 on / type reiser4 >(rw,iocharset=utf8,optimal_io_size=2097152,atom_max_size=0x800) > > > > >Edward Shishkin wrote: > > >>butdiene wrote: >> >> >> >>>right now, I'm using the following setting. mount options are not the same >>> >>> >as > > >>>in /proc/mounts, I'm not sure which ones are accepted. >>> >>>but the system seems to be more responsible now. the optimal_io_size was >>>64KB, seems to be too small. >>> >>>I reduced the atom_max_size (from 0x2f238 = 784MB) to 0x800= 8MB. >>> >>>not sure how they would help. I will report later if it really helps. >>> >>>thanks >>> >>>mount|grep sda3;cat /proc/mounts |grep root >>>/dev/sda3 on / type reiser4 >>>(rw,iocharset=utf8,tmgr.atom_max_flushers=0,optimal_io_size=2097152,atom_max_size=0x800,atom_max_age=0x600,atom_max_flushers=0) >>>rootfs / rootfs rw 0 0 >>>/dev/root / reiser4 >>> >>> >>> >>> >>atom_max_size=0x2f238,atom_max_age=0x927c0,atom_min_size=0x100,atom_max_flushers=0x1,cbk_cache_slots=0x10 >> >>^^^^^^^^^^^^^^^^^^^^ >> >>You have atom_max_size unchanged. >>Please, edit your /etc/fstab and reboot. >> >>Thanks, >>Edward. >> >> >> >>>0 0 >>> >>> >>>Edward Shishkin wrote: >>> >>> >>> >>> >>>>butdiene wrote: >>>> >>>> >>>> >>>> >>>> >>>>>Hi, >>>>> >>>>>I run reiser4 on a laptop. I got serious fsync/flush problem. >>>>> >>>>>almost randomly, the hard drive would run busy for minutes to dozens of >>>>>minutes. while the computer is still responsive during this period, but >>>>>firefox would hang until the hard gets idle again. >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>Hm.. firefox is fsync-bound app? >>>> >>>> >>>> >>>> >>>> >>>>>I guess I can set mount options to improve the situation. >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>That hang can have a complex nature. >>>> >>>>Well, try to play with the mount option "tmgr.atom_max_size=N", >>>>where N is atom size in blocks (by default it is total RAM pages / 4). >>>>Look at /proc/mounts to make sure that your changes take effect. >>>> >>>>If it doesn't help, then let me know. >>>> >>>>Thanks, >>>>Edward. >>>> >>>> >>>> >>>> >>>> >>>>>Any suggestion is appreciated. >>>>> >>>>>thanks >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>- >>>>To unsubscribe from this list: send the line "unsubscribe reiserfs-devel" >>>>in >>>>the body of a message to majordomo@vger.kernel.org >>>>More majordomo info at http://vger.kernel.org/majordomo-info.html >>>> >>>> >>>> >>>> >>>> >>>> >>> >>> >>> >>> >>- >>To unsubscribe from this list: send the line "unsubscribe reiserfs-devel" >>in >>the body of a message to majordomo@vger.kernel.org >>More majordomo info at http://vger.kernel.org/majordomo-info.html >> >> >> >> > > >