Date: Tue, 28 Jul 2015 10:28:39 +0200 From: Matthias Apitz <guru@unixarea.de> To: David Chisnall <theraven@FreeBSD.org> Cc: Slawa Olhovchenkov <slw@zxy.spb.ru>, freebsd-current@freebsd.org Subject: Re: duration of buildworld Message-ID: <20150728082839.GA2521@c720-r276659> In-Reply-To: <8E86BFED-FE49-495D-B2F4-B6D2F0CB9DFA@FreeBSD.org> References: <20150727055804.GA2312@c720-r276659> <20150727120006.GJ43740@zxy.spb.ru> <20150727184704.GA1777@c720-r276659> <20150727193410.GX44094@zxy.spb.ru> <20150727194039.GA2298@c720-r276659> <20150727194954.GY44094@zxy.spb.ru> <8E86BFED-FE49-495D-B2F4-B6D2F0CB9DFA@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
El día Tuesday, July 28, 2015 a las 09:20:38AM +0100, David Chisnall escribió: > It sounds as if the swap was working (as the build took a long time, it didn’t run out of memory). My guess would be that, before the reboot, something had wired (or, if not, then touched frequently enough to keep it in core) a large chunk of memory, which forced the swapping. Swap files are expected to be slower than swap partitions (though if you’re swapping to a disk, the seek times are likely to dominate in both cases). > > I don’t suppose there’s capture of the output of top from before the reboot? No, I did not saved it. When I started the buildworld, I started in parallel already a poudriere jail to build around 1600 ports. Later, when I realized that the buildworld took so long, I stopped (killed) the poudriere jail. IIRC the "swapctl -l" showed before the boot that the swap files have been used. I could re-create the situation on next buildworld. matthias -- Matthias Apitz, ✉ guru@unixarea.de, http://www.unixarea.de/ ☎ +49-176-38902045 No! Nein! ¡No! Όχι! -- Ευχαριστούμε!
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20150728082839.GA2521>