This information is based upon bug reports against RedHat and Ubuntu. Even though there is a claim that these are fixed in recent kernel versions, we (and other users) have still observed this error, so the issue may not be resolved all the way. The recommended versions have shown significantly reduced occurrences in our testing.
- RHEL versions later than 2.6.18-194.32 and earlier than 2.6.32-131
- Ubuntu versions 2.6.31, 32, 33, and 34
Affected systems may encounter reduced performance on both ext3 and ext4 filesystems. Symptoms may include:
- hung_task_timeout_secs messages in system dmesg logs
- widely variable agent availability (entering and exiting agent penalty status frequently)
- contention over the ecagent.state file
- slower builds (with unexplained variances)
To help determine if your environment has this kernel issue, rundmesg | grep hung_task_timeoutIf hung_task_timeout errors are present, this indicates that a known Linux kernel issue is present. Contact your kernel provider to obtain a different version of the precompiled kernel.
The issue appears to be related to heavy I/O load.
The effect appears to occur after some period of time, so rebooting the agents on a regular basis may help.
Consider upgrading to 2.6.32-131 (RHEL 6.1), or downgrading to 2.6.18-194.32 (RHEL 5.5).
Consider upgrading to kernel version 2.6.35 or later.
To install the upstream kernel, do the following, for example:
Upgrading the kernel may require you to reinstall various kernel modules such as video drivers, efs, and vm-tools. You can do so by running: apt-get dist-upgrade and then rebooting.
Additional information can be found here.
Solaris 9 support for Cluster Manager machines ended as of Accelerator v6.1. Because Accelerator v6.1 uses MySQL 5.5, and MySQL 5.5 does not support Solaris 9, Accelerator no longer supports Solaris 9 as a Cluster Manager platform.
Due to missing kernel-level APIs, ElectricAccelerator does not support registry virtualization on this platform.