ssh daemon hangs after publickey packet sent

Bug #613022 reported by Scott Moser on 2010-08-03
This bug affects 3 people
Affects Status Importance Assigned to Milestone
cloud-init (Ubuntu)
linux (Ubuntu)
openssh (Ubuntu)

Bug Description

A launched ec2 instance in ap-southeast-1 region is unreachable via ssh.
$ ssh -vvv
shows progress up to :

debug3: authmethod_is_enabled publickey
debug1: Next authentication method: publickey
debug1: Offering public key: smoser@brickies
debug3: send_pubkey_test
debug2: we sent a publickey packet, wait for reply

Then nothing for minutes before session is killed (manually).

In a 'good' connection, the following would be next:
debug2: we sent a publickey packet, wait for reply
debug1: Authentications that can continue: publickey

I'll attach full 'ssh -vvv' logs good and bad connection.

ProblemType: Bug
DistroRelease: Ubuntu 10.10
Package: openssh-server 1:5.5p1-4ubuntu3
ProcVersionSignature: User Name 2.6.35-14.19-virtual 2.6.35
Uname: Linux 2.6.35-14-virtual x86_64
Architecture: amd64
Date: Tue Aug 3 14:45:25 2010
Ec2AMI: ami-9fc4bbcd
Ec2AMIManifest: ubuntu-images-testing-ap-southeast-1/ubuntu-maverick-daily-amd64-server-20100803.1.manifest.xml
Ec2AvailabilityZone: ap-southeast-1a
Ec2InstanceType: m1.large
Ec2Kernel: aki-11d5aa43
Ec2Ramdisk: unavailable
SourcePackage: openssh

Scott Moser (smoser) wrote :
Scott Moser (smoser) wrote :
Scott Moser (smoser) wrote :
Scott Moser (smoser) wrote :

Other information:
ssh-keyscan did show results:

# SSH-2.0-OpenSSH_5.5p1 Debian-4ubuntu3 ssh-rsa AAAAB3NzaC1y...RzIcF9UuISZaYJf

A reboot-instances seems to have no affect.

What is strange here is that there was a key in ssh-keyscan, yet the console shows no key-gen output. This is likely related to the fact that the instance would not reboot.

I suspect kernel problem.

affects: linux-meta (Ubuntu) → linux (Ubuntu)
Scott Moser (smoser) wrote :

The reboot-instances did eventually occur (it did not have immediate effect), and after, it came up fine.
I'm attaching the post-reboot log here.

It contains messages like:
[ 240.170158] INFO: task sshd:668 blocked for more than 120 seconds.^M
[ 240.170175] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.^M
[ 360.170086] INFO: task sshd:668 blocked for more than 120 seconds.^M
[ 360.170102] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.^M

Scott Moser (smoser) wrote :
Scott Moser (smoser) wrote :

I've seen this again now, on:
eu-west-1 ami-53dbf127 canonical ubuntu-maverick-daily-amd64-server-20100803.1

Mathias Gug (mathiaz) on 2010-08-03
Changed in linux (Ubuntu):
importance: Undecided → Medium
Changed in cloud-init (Ubuntu):
importance: Undecided → Medium
Changed in openssh (Ubuntu):
importance: Undecided → Medium
tags: added: iso-testing
tags: added: review-request
Scott Moser (smoser) wrote :

Marking this wishlist, it appears very uncommon and i really think its either ssh client or server that has the issue.

Changed in cloud-init (Ubuntu):
importance: Medium → Wishlist
status: New → Confirmed
Brad Figg (brad-figg) on 2011-02-09
Changed in linux (Ubuntu):
importance: Medium → Undecided
status: New → Invalid
Ben Howard (utlemming) wrote :

Scott -- are the instances that exhibit this bug EBS-backed or S3-backed?

Scott Moser (smoser) wrote :

Well, the two ami ids listed above (in the bug opening and then comment 7) were both instance-store.

Scott Moser (smoser) on 2011-12-15
Changed in openssh (Ubuntu):
status: New → Confirmed
Scott Moser (smoser) wrote :

marking invalid on cloud-init, definitely the bug was not/is-not here.

Changed in cloud-init (Ubuntu):
status: Confirmed → Invalid
To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers