httpd-2.4.23 notice "No slotmem from mod_heartmonitor"
Thu Aug 4 23:43:00 GMT 2016
>> Activating mod_slotmem_shm (as a shared module) resolved this issue and
>> allowed httpd to start. However, httpd now produces the following notice
>> on (successful) startup:
>> [Mon Jul 25 09:27:40.718810 2016] [lbmethod_heartbeat:notice] [pid2345]
>> AH02282: No slotmem from mod_heartmonitor
>> I'm wondering if there changes to the static modules included in the new
>> version's package. The changelog for httpd itself didn't suggest any changes
>> between these two versions that would cause configuration differences.
I was able to eliminate the notice by adding --disable-lbmethod-heartbeat to CYGCONF_ARGS in the .cygport file. Looking up the notice in the package's source code, it appears that the lbmethod_heartbeat module is attempting to use shared memory (slotmem) created by the heartmonitor module (using the slotmem_shm module) and failing, thus producing the notice. While this notice appears to be nonfatal, I wonder if it would pose a problem if actually using proxy load balancing; the lbmethod_heartbeat module provides one of four algorithms available for proxy load balancing and may not work properly.
Interestingly, this issue does not appear under the previous version, 2.4.20, of the httpd package. Comparing the .cygport files for 2.4.20 and 2.4.23 shows that the only difference is the value of VERSION. Also, 2.4.23 now requires the slotmem_shm module (either static or shared) whereas 2.4.20 did not. It appears that there may have been an upstream change resulting in this issue. However, reviewing the changelog for httpd itself (versions 2.4.21 through 2.4.23) didn't reveal any apparently relevant changes.
Experimenting with my previous thought that other modules may need to be compiled as static, instead of shared, modules didn't resolve the issue.
As noted above, disabling the lbmethod_heartbeat module removes the notice. This action doesn't address the root problem for someone who wants to use this proxy load balancing algorithm, but it should prevent accidental use of an algorithm that may be functioning improperly.
Dominic R. Jones, Ph.D.
Problem reports: http://cygwin.com/problems.html
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
More information about the Cygwin