summary refs log tree commit diff stats
path: root/results/classifier/gemma3:12b/files/1502095
blob: 956d8e5a39673c77fccf5108bbd24b6813b4e9c7 (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
Sporadic input / output error — x86-64 linux guest

** Setup: **

→ Host
    Qemu version 2.4.0.1
    Linux: 4.1.1 (Debian 8.2, GCC 4.9.2, x86_64)
    filesystem ext3 and ext4

→ Guests (3 VMs)
    architecture x86_64, Linux 3.16.0-4-amd64 (Debian 7.6)
    virtual disk qcow2, uncompressed
    guests filesystem ext3
    virtual disks size:  VM1: 3GB,  VM2: 5GB,  VM3: 250GB

→ Network
    bridge (br0) and tap interfaces.


** Command line **

For all 3 VMs, the command line is similar to the following. Only RAM size and ID ("109") are changed.

    /usr/local/bin/qemu-system-x86_64 -hda /media/raid1/qemu-109 -m 8G -smp 4 -enable-kvm \
        -netdev bridge,id=br109 -device virtio-net-pci,netdev=br109,id=nic0,mac=00:00:00:00:01:09 \
        -k it -daemonize


** Problem **

Sporadic, unexplained input output error.
When I try to SSH to one of the instances, most of the times everything just works fine. Some other times, the SSH connection can be established, but as I interact with the guests via SSH (ls, cd, top) the guest reports "input / output error" on the SSH console. Sometimes, the SSH daemon on the guest answers "/bin/bash input output error". Sometimes the SSH client answers that the connection has been dropped by the host.

When this happens, the web services I run on the VMs get unreachable, the VMs themselves are unreachable/unusable via SSH as described, and after killing the relative qemu processes and restarting the VMs, no recent logs are registered on /var/logs, arguably since the time of the crash.

This only happens to one VM at a time, independently. The other VMs appear to run fine when one VM encounters the problem.

I which instructions to further debug this.