On Sat, Feb 22, 2025 at 08:52:46AM -0800, Anthony Sorace wrote: > I think you?re misunderstanding that 10 second wake up process. Dumps are not > taken every 10 seconds, but once a day. See this part of the paper, slightly > above: > > The Cwrite and Cdirty blocks are created and never removed. Unless something > is done to convert these blocks, the c-device will gradually fill up and stop > functioning. Once a day, or by command, a dump of the cw-device is taken. The > purpose of a dump is to queue the writes that have been shunted to the > c-device to be written to the w-device. Since the w-device is a WORM, blocks > cannot be rewritten. Blocks that have already been written to the WORM must > be relocated to the unused portion of the w-device. These are precisely the > blocks withCwrite state. > > The process that wakes every 10 seconds is for walking through the queue of > blocks that are to be written to the worm, but it doesn?t do anything to look > at files you are working with; it doesn?t add anything to that queue. That > only happens once a day. >
Oh! OK. But this means that if there is a memory constraint, the kernel has no other choice than committing the buffer cache to gain room, and in this case there will be intermediary (in-between dumps, deliberate of forced by an halt) to be committed to the WORM? (If the editing of a modified file is idle, and there is more modifications done on active files.) So scp and wcp are triggered by dump? > > On Feb 22, 2025, at 04:15, tlaro...@kergis.com wrote: > > > > ?WORM is a great idea, allowing to store only the diff and to be > > able to version is a bonus. > > > > But there is one drawback: when one is modifying rapidly a file > > (tentatives), it can spoil the worm with useless modifications. > > > > The fossil snapshots were a solution (allowing to "commit" on demand). > > > > From https://9p.io/sys/doc/fs/fs.html, the copying of modifications > > from fscache to buffer (scp) as well as the copying of the buffer > > to the WORM (wcp) is done every ten seconds. > > > > Is there a way, from userland, to modify the frequency of the > > scp and wcp depending on the binding name? > > > > Has someone attempted to build a stack of other and fscache, so that > > the writable upper level (other) is used as a chalkboard and, when the > > user is satisfied with the tentative file, he can "commit" this > > "minute" (sketch) to the fscache---letting then the normal procedures > > register the differences in the WORM? > > > > Related: when using git, git by itself has the registration of the > > modifications. So the .git could be put in fscache, but the current > > copy should be in other. Is there a way to achieve this---if I > > understand correctly (I may not), gitfs serves the .git hierarchy but > > is not creating as an artefact the working directory, making the > > branch appear, and overwriting with locally modified not committed > > files? (Hoping my sort of English is sufficiently understandable...) > > -- > > Thierry Laronde <tlaronde +AT+ kergis +dot+ com> > > http://www.kergis.com/ > > http://kertex.kergis.com/ > > Key fingerprint = 0FF7 E906 FBAF FE95 FD89 250D 52B1 AE95 6006 F40C -- Thierry Laronde <tlaronde +AT+ kergis +dot+ com> http://www.kergis.com/ http://kertex.kergis.com/ Key fingerprint = 0FF7 E906 FBAF FE95 FD89 250D 52B1 AE95 6006 F40C ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T5a27390f138b4be1-Me02925c8d6fcdf8de2012ba2 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
