Jsmart
12-01-2009, 03:44 PM
hf3389 is a mandatory hotfix for AppLogic 2.7.8 and cannot be used with other versions of AppLogic. The hotfix installs updated AppLogic binaries on the servers and grid controller. The /system:WEB5 and /system:WEB64 boot volumes are also modified.
A grid reboot is required after applying the hotfix. However, a majority of the fixes only require a grid controller reboot (which in AppLogic 2.7 does not cause application downtime). If the running applications on your grid are sensitive to downtime, it is recommended to reboot the grid controller immediately after installing this hotfix. Rebooting the full grid (all servers and applications) can be delayed until a later time such as a scheduled maintenance window (the fixes that require a grid reboot to take effect are very rare bugs that many users are unlikely to encounter).
3tera recommends to install this hotfix on all AppLogic 2.7.8 grids.
This hotfix resolves several issues with AppLogic 2.7.8. These include fixes for the following defects:
* 2488: user cannot login or view the text console for legacy appliances (ssh and text console access are disabled)
* 2824: info srv --map shows the controller as being on the wrong server
* 3320: server is not rebooted if its last known state was down and it connects to the controller with allocated resources
* 3355: branching an assembly class leads to invalid/unusable branched assembly
* 3368: WEB5/WEB64 appliance does not execute .rc.local on boot
* 3370: component list --extended --batch may display incorrect component info
* 3374: app provision fails if memory or volume size is specified as a fractional value containing 0 (i.e., 128.0)
* 3379: grid controller recovery: controller may be restarted twice as a result of a flapping controller server
* 3383: an appliance that crashes on boot may not be restarted (intermittent)
* 3384: assigning a new primary server may fail (rare)
* 3385: resources may not be cleaned up when stopping an appliance with comp stop --force
* 3386: an incorrect number of failed application recoveries may be reported after a grid recovery has been aborted
* 3387: log a dashboard message if an empty server reports allocated resources
* 3388: disable storage failure detection
This hotfix also provides workarounds for the following defects:
* 2074: occasionally the grid controller fails due to an oom killer (rare)
* 3381: stuck app stop command resulted in an incomplete grid recovery after a controller failure
A grid reboot is required after applying the hotfix. However, a majority of the fixes only require a grid controller reboot (which in AppLogic 2.7 does not cause application downtime). If the running applications on your grid are sensitive to downtime, it is recommended to reboot the grid controller immediately after installing this hotfix. Rebooting the full grid (all servers and applications) can be delayed until a later time such as a scheduled maintenance window (the fixes that require a grid reboot to take effect are very rare bugs that many users are unlikely to encounter).
3tera recommends to install this hotfix on all AppLogic 2.7.8 grids.
This hotfix resolves several issues with AppLogic 2.7.8. These include fixes for the following defects:
* 2488: user cannot login or view the text console for legacy appliances (ssh and text console access are disabled)
* 2824: info srv --map shows the controller as being on the wrong server
* 3320: server is not rebooted if its last known state was down and it connects to the controller with allocated resources
* 3355: branching an assembly class leads to invalid/unusable branched assembly
* 3368: WEB5/WEB64 appliance does not execute .rc.local on boot
* 3370: component list --extended --batch may display incorrect component info
* 3374: app provision fails if memory or volume size is specified as a fractional value containing 0 (i.e., 128.0)
* 3379: grid controller recovery: controller may be restarted twice as a result of a flapping controller server
* 3383: an appliance that crashes on boot may not be restarted (intermittent)
* 3384: assigning a new primary server may fail (rare)
* 3385: resources may not be cleaned up when stopping an appliance with comp stop --force
* 3386: an incorrect number of failed application recoveries may be reported after a grid recovery has been aborted
* 3387: log a dashboard message if an empty server reports allocated resources
* 3388: disable storage failure detection
This hotfix also provides workarounds for the following defects:
* 2074: occasionally the grid controller fails due to an oom killer (rare)
* 3381: stuck app stop command resulted in an incomplete grid recovery after a controller failure