Hi Dan, I can see that this issue is becoming more nuanced. No files were moved and it is extremely unlikely that disk corruption is involved. gfsh reports that a disk store is missing and “revoke missing-disk-stores” alleviates the problem. However, I am also using the HTTP Session Replication Module, which I just learned has its own embedded cache.xml (http://gemfire.docs.pivotal.io/latest/tools_modules/http_session_mgmt/weblogic_changing_gf_default_cfg.html <http://gemfire.docs.pivotal.io/latest/tools_modules/http_session_mgmt/weblogic_changing_gf_default_cfg.html>).
The original thought was to share a single cluster between user-defined regions and the http session module. The cache.xml for the user-defined regions has a disk-store defined. I suspect that the http session module cache.xml may also... the region is defined as persistent. My working hypothesis at present is that the two definitions are colliding with one another somehow? The repeatable problem occurs when: 1) cluster starts 2) tomcat starts 3) cluster bounces - “missing disk store” - missing for the sessions region but not the user region? I am investigating but I will need to coordinate with others to get the http session module cache.xml. Do you know if a single cluster can be shared between the http session module cache.xml and a different set of xml? The regions specified in each get created. Thank you for the reply! > On Sep 30, 2015, at 1:41 PM, Dan Smith <[email protected]> wrote: > > Hi Wes, > > A few questions: > 1) Are you sure your disk store is actually there - in the same location > as reported by list missing disk stores? > 2) Are you sure you've started all members and the members are still > waiting? Use list members and list disk missing disk stores in gfsh to > confirm. > 3) Did you move or delete files from that location at some point? Even if > gemfire recreates files in the same location, it will still report the old > files as missing, because the disk store is a assigned a unique id which is > what geode really cares about. > > -Dan > > On Wed, Sep 30, 2015 at 10:29 AM, Real Wes Williams <[email protected]> > wrote: > >> >> Page 779 of the User's Guide at >> http://gemfire.docs.pivotal.io/pdf/pivotal-gemfire-ug.pdf < >> http://gemfire.docs.pivotal.io/pdf/pivotal-gemfire-ug.pdf> describes the >> administrative procedure "revoke missing-disk-store” to cleanly proceed >> when there is a missing disk store. >> >> I am experiencing this but there are disk stores in the cache server on >> which other cache servers are waiting. All I did was a shutdown via gfsh >> and restarted using a new jar and a cache.xml function name change. >> >> Why do disk store’s report as missing when they are ostensibly not >> missing ?? >> >> -rw-r--r--. 1 gemfire pivotal 1829 Sep 28 15:49 BACKUPDEFAULT_3.crf >> -rw-r--r--. 1 gemfire pivotal 260 Sep 28 15:49 BACKUPDEFAULT_3.drf >> -rw-r--r--. 1 gemfire pivotal 1758 Sep 23 12:50 BACKUPDEFAULT_3.krf >> -rw-r--r--. 1 gemfire pivotal 2951 Sep 29 13:18 BACKUPDEFAULT_5.crf >> -rw-r--r--. 1 gemfire pivotal 260 Sep 29 13:18 BACKUPDEFAULT_5.drf >> -rw-r--r--. 1 gemfire pivotal 5242880 Sep 28 15:49 BACKUPDEFAULT.if >> >> >>
