Bug#783063: Xen domU freeze with "Guest Rx stalled"
On Tue, 2015-04-21 at 13:31 +0200, Wolodja Wentland wrote:
> Package: src:linux
> Version: 3.16.7-ckt4-3~bpo70+1
> Severity: important
>
> Hello,
>
> ever since upgrading some of our Xen dom0s to the kernel in backports at
> that time (3.16.7-ckt4-3~bpo70+1) we are seeing domU freezes with output such
> as:
>
> [2951591.712865] vif vif-1-0 vif1.0: Guest Rx stalled
> [2951591.713145] public: port 2(vif1.0) entered disabled state
>
> and a normal shutdown fails with:
>
> [2015-04-21 11:24:03 3895] DEBUG (XendDomainInfo:1881) XendDomainInfo.handleShutdownWatch
> [2015-04-21 11:27:02 3895] INFO (XendDomainInfo:2114) Domain shutdown timeout expired: name=foo-domU1 id=1
> [2015-04-21 11:28:02 3895] DEBUG (XendDomainInfo:524) XendDomainInfo.shutdown(poweroff)
OOI what is your dom0 environment/versions etc?
>
> so that we had to destroy it for it to (re-)boot.
>
> It should be noted that we have not yet seen this problem on PVHVM guests, but
> only on PV ones. The domains are shown as running (state r), but do not react to
> any commands.
It seems like it has deadlocked, but I don't know if that is a Xen issue
or some generic issue.
> A patch that *might* be related is xen-netback-reintroduce-guest-Rx-stall-detection.patch
> (cf. [0]), but we are not sure about that yet.
This should be in the kernel you are running already, it was added in
3.16.7-ckt2.
Anyway, bpo now contains 3.16.7-ckt7-1~bpo70+1 which includes quite a
few stable updates over the version you are running. Please could you
try that.
There is a fair bit of stuff only in jessie in the 3.16.7-ckt9-2. Not
sure when that is due to hit bpo though.
Ian.
Reply to: