Date: Fri, 25 Jan 2002 13:53:03 -0800 From: Terry Lambert <tlambert2@mindspring.com> To: Brad Knowles <brad.knowles@skynet.be> Cc: Mike Meyer <mwm-dated-1012390758.50933b@mired.org>, chip <chip@wiegand.org>, freebsd-chat@freebsd.org Subject: Re: Bad disk partitioning policies (was: "Re: FreeBSD Intaller (was "Re: ... RedHat ...")") Message-ID: <3C51D3BF.FB72A84C@mindspring.com> References: <20020123114658.A514@lpt.ens.fr> <20020123124025.A60889@HAL9000.wox.org> <3C4F5BEE.294FDCF5@mindspring.com> <20020123223104.SM01952@there> <p0510122eb875d9456cf4@[10.0.1.3]> <15440.35155.637495.417404@guru.mired.org> <p0510123fb876493753e0@[10.0.1.3]> <15440.53202.747536.126815@guru.mired.org> <p05101242b876db6cd5d7@[10.0.1.3]> <15441.17382.77737.291074@guru.mired.org> <p05101245b8771d04e19b@[10.0.1.3]>
next in thread | previous in thread | raw e-mail | index | archive | help
There is a fundamental problem with /var. Having /var/run and /var/log on the same partition means that if your log files fill up, you will be unable to start some programs. Having /var/log on the same parititon as TMPDIR (for whatever program, and by whatever means... e.g. environment, sendmail.cf, or a symbolic link of /tmp to /var/tmp, etc.) means that if you fill up your log files, then you fill up your /tmp directory. For email, this means no more email received, and other programs have similar bad behaviour. Historically, I have seen this very often with programs that rely on these resources; sendmail has been a particular problem, though not because of any fault on its part: it's just that email is about the most visible service failure you can have with any server, no matter what else you run on it. My particular problems (in a commercial product, which sold, or leased, tens of thousands of units) was that "newsyslog" would not recreate history when it failed and then was rerun, and "samba" log files are hard to rotate using the standard means. Specifically, if cron dies (common, particularly if your password file is read-only, due to an infrequent VM bug having to do with copy-on-write of modified read-only pages from the mmap of the passwd.db files, since "cron" rewrites the password entry from getpwent(3) in place, making it dirty, by its assumption that the pointers point to a static buffer in libc, not a memory mapped region in a read-only file), then newsyslog stops being run by cron. When that happens, log files become huge, and then when newsyslog is restarted, it only "rolls the files over" by one -- leaving the huge accumulated log files as the one-behind ...and incidently, leaving /var/log *still full*. In any case, the point of this is to note that, no matter how you slice it, there is going to be a monster in the closet, waiting for your mother to turn out the light and close the door. -- Terry To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-chat" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3C51D3BF.FB72A84C>