Date: Thu, 16 Apr 2020 23:52:52 +0700 From: Eugene Grosbein <eugen@grosbein.net> To: Wojciech Puchar <wojtek@puchar.net> Cc: freebsd-hackers@freebsd.org, Poul-Henning Kamp <phk@phk.freebsd.dk> Subject: Re: fsck parallel check Message-ID: <571d8b81-f4f9-d371-6cf2-2e7aeb82a09b@grosbein.net> In-Reply-To: <alpine.BSF.2.20.2004161824360.3899@puchar.net> References: <alpine.BSF.2.20.2004161748390.20589@puchar.net> <e5742016-81bb-2ba3-74a4-c633d3627b23@grosbein.net> <alpine.BSF.2.20.2004161824360.3899@puchar.net>
next in thread | previous in thread | raw e-mail | index | archive | help
16.04.2020 23:25, Wojciech Puchar wrote: >> Quoting the manual: >> >> In other words: In preen mode all pass 1 partitions are checked >> sequentially. Next all pass 2 partitions are checked in parallel, one >> process per disk drive. Next all pass 3 partitions are checked in >> parallel, one process per disk drive. etc. > > Yes i've read this manual. > > > "Next all pass 2 partitions are checked in parallel, one > process per disk drive." > > so having 5 partitions on one SSD - it will be checked sequentially. Not necessary. Fsck tries to parse device name this way: /path/to/nameNUMBERrest so it skips all chars upto last /, then extracts "name" part not containing any digits, then extracts NUMBER part containing only digits, upto maximum lenght. Then is uses nameNUMBER for "disk name". So if your naming schema is /dev/gpt/gptlabel it all depends on exact label names because /dev/gpt/ part is skipped first. If your gptlabels are in form of nameNUMBERrest where name and NUMBER are all same, fsck threats them as part of single disk and scans sequentially. Otherwise, in parallel. You may easily check it with "fsck -d -p" so it prints verbose details not starting any checks really.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?571d8b81-f4f9-d371-6cf2-2e7aeb82a09b>