Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On 2011-05-05 21:47, Jeff Humby wrote: > Has anyone achieved success with loopback devices? Any suggestions Suppose you have a filesystem on top of DRBD and that DRBD is a loop device in another filesystem, and you're getting memory pressure. Any process: Memory, this pid whatever. Request a page. Memory allocator (malloc): pid whatever, copy, stand by. malloc: kernel threads, this is malloc, perform writeout. Filesystem: roger that malloc, stand by. Filesystem: DRBD, write this page. DRBD: 10-4, writing page (no need to talk to malloc, we have a mempool). Stand by. DRBD: loop driver, here's a page, write it. Loop driver: Roger that DRBD, processing. Loop driver: filesystem, here's a page you need to write Filesystem: Roger that loop. Filesystem: malloc, I need some memory. malloc: copy that Filesystem, stand by until we free some pages. Filesystem: malloc, are you telling me to bugger off? You just caused a writeout! Do you expect pages to be freed magically? malloc: Affirmative. Filesystem: Damn. Loop driver: Damn. DRBD: Damn. Process which requested page: Damn. See my point? Don't put DRBD on loop devices. Cheers, Florian -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 262 bytes Desc: OpenPGP digital signature URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20110506/d55891aa/attachment.pgp>