Shared Folders folder disappeared
Hi all
We've had the strangest thing happen! The Shared Folders folder disappeared from the catalog. It has happened twice now, on two different environments. We've checked all permissions and apart from the small group of Admins, there is no one who could've deleted it.
There don't seem to be an easy way to see who deleted it. That is if it was actually deleted - we are not convinced yet that it was a deliberate act. Is there any system process or known bug that could cause the Shared Folder to disappear?
Thanks
J
Answers
-
Ok obvious first question: have you actually checked the physical file system?
Have you checked the .atr file which contains the REAL access rights in terms of the OBIPS? You can simply check what's written in there.
0 -
Hi J,
Is it really deleted? Have you checked the catalog on the OS?
As far as I know, there is no process that automatically deletes the Shared Folders [quite happy about that ]
Have your reviewed this blog by Robin?
Michael
0 -
It would be helpful to know the version and bundle patch as well.
0 -
@SteveF-Oracle are you new here of just still optimistic enough that you dare ask this? ;-)
0 -
@user11394838 aside from the other questions - what OS are you on? Single or multi-node cluster?
0 -
lol, optimistic I suppose. I do wish one could force a template for basic information vs. 20 questions game. The other alternative is ignore, I suppose, but that is no fun.
I've come to love the http://catb.org/~esr/faqs/smart-questions.html#beprecise that I often see posted by @rmoff
0 -
-
Thanks for the feedback and apologies for the missing info.
This is OBIEE 11.1.1.7.141014 on an Exalytics T5-8 system R1 patchset 5. The OS is Solaris 11.1 and it is a two node cluster.
Yes, we checked the physical file system. Under the root folder the Shared folder and the shared.atr file are gone. Only the System and Users folders and their .atr files are still there.
0 -
I would still go for the blog of Robin and check the logging:
0 -
Thanks Michael, I did read Robin's blog before I posted here. We have a system freeze until end of the month so I can only implement that next month.
The system owners are understandably quite nervous as this has happened twice in 3 weeks. Other than a human deliberately deleted the folder, are there any other possible explanations?
0 -
Well if it wasn't a human then .. it was a computer
And OBIEE wouldn't have got to where it is today if it made a habit of soiling its pants like this, so I would imagine it's something else running that's doing it (if it's not a human - malicious/accidental/ignorant). Can you speak to your Solaris admins and see what housekeeping (maintenance, backup etc) jobs run?
How are you sharing the webcat, is it a NFS mount between the two nodes? Two physical boxes or one partitioned?
The only time I've read about webcats getting seriously fubar'd is when there's clustering and non-local FS involved.
I'd be inclined to also get an SR raised at this point, if only to start on the process of diagnostics from a second angle (and the one that you pay your support for), in case it does turn out to be a bug. That you're on an engineered system gives added credence to the point that it should Just Work...
0 -
Thanks for the feedback Robin!
The Webcat and RPD is indeed on an NFS shared between the two nodes. Two boxes, a node on each.
I checked the file system on both nodes and the catalog folder only appears on the one node, and it is there where it disappeared from. Solaris admin says there is nothing running that could cause this problem - the only jobs are log maintenance jobs.
We had to immediately restore from backup without there being an opportunity for proper troubleshooting. If - IF - this happens again, an SR will be the first thing I do.
Thanks again
J
0 -
the catalog folder only appears on the one node, and it is there where it disappeared from.
OK I'm confused again
Is this a single BI domain, clustered across two nodes? If so, how do you mean it disappeared from *one* of the nodes, and that it only appears on *one* node?
Normally you'd have the shared network folder mounted on each BI host. Otherwise, how do you manage the catalog within the instance?
0 -
The <CatalogPath> in instanceconfig points to a Solaris mount point to the NFS.
If I browse the file system, I can see the location /u01/app/oracle/product/fmw/instances/instance1/bifoundation/OracleBIPresentationServicesComponent/coreapplication_obips1/catalog exists on node 1. This location does not exist on node 2. So essentially the shared folder only exists on node 1 and it disappeared from node 1, rendering the system unusable.
Hope that makes sense?
0 -
1. Is this a single BI domain, clustered across two nodes?
2. On node 2, what is CatalogPath set to in instanceconfig.xml?
3. Is the folder mounted on each node at the same filesystem point?
0 -
This really starting to sound weird.
0
