Date: Thu, 3 Nov 2016 23:54:46 +0100 From: Rainer Duffner <rainer@ultra-secure.de> To: FreeBSD Filesystems <freebsd-fs@freebsd.org> Cc: freebsd-stable <freebsd-stable@freebsd.org> Subject: Re: zfs receive leads to a stuck server Message-ID: <60B441B0-F5C8-417D-95CF-23ED52E252F0@ultra-secure.de> In-Reply-To: <4707908B-4868-4AA6-ADD6-D24121EFAE38@ultra-secure.de> References: <4707908B-4868-4AA6-ADD6-D24121EFAE38@ultra-secure.de>
next in thread | previous in thread | raw e-mail | index | archive | help
> Am 01.10.2016 um 18:02 schrieb Rainer Duffner = <rainer@ultra-secure.de>: >=20 > Hi, >=20 > I posted this before, but I didn=E2=80=99t really get an answer and = I=E2=80=99m still looking for ways to debug this. >=20 > I have to servers (HP DL 380 Gen8, 192 GB RAM, 6C CPU). > I=E2=80=99ve outfitted them with HP=E2=80=99s H22x cards (really OEMed = 9207-8x, three altogether) that I recently cross-flashed to LSI=E2=80=99s = latest firmware. And it turns out, using anything branded by HP is just asking for ulcers = and headaches. We replaced said HBAs with LSI^WAvago^WBroadcom HBAs - and everything = started working as it should. There are still hangs at 03:00 and 04:00 (when the snapshots that got = deleted on the master get deleted on the slave) - but that's not really = a problem because they are much shorter than before and nobody is using = the system at that time. So, while the H22x-cards are =E2=80=9Eclose enough to a LSI2308-based = card so that the driver actually thinks it=E2=80=99s a 2308 - they are = in fact not. Also, with the original LSI cards, you can use LSI=E2=80=99s FreeBSD = flash utility and have a far better life overall.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?60B441B0-F5C8-417D-95CF-23ED52E252F0>