This post is more than 5 years old
1 Rookie
•
64 Posts
1
2950
August 18th, 2017 10:00
CloudPools and OneFS 8.0.0.4
I'm having issues with trying to manually archive to a CloudPool. I'm OneFS 8.0.0.4 and don't see any specific patches that might address this. Planning to open a call next week but looking for input before I do.
I am connecting to Azure and everything appears correct with my Cloud Storage Account (enabled) and my CloudPool (enabled). I am fully licensed for SmartPools and CloudPools. However, when I run a test to manually archive, nothing gets put in the cloud (it never stubs):
isi cloud archive /ifs/isi01/cptest/pic1.jpg -v
isi cloud job view [job number]
I see the following:
ID: 316
Description:
Effective State: completed
Type: archive
Operation State: running
Job State: completed
Create Time: 2017-08-18T09:54:16
State Change Time: 2017-08-18T09:54:17
Completion Time: 2017-08-18T09:54:17
Job Engine Job: 8408
Job Engine State: succeeded
Total Files: 0
Total Canceled: 0
Total Failed: 0
Total Pending: 0
Total Staged: 0
Total Processing: 0
Total Succeeded: 0
When I dig into the log (/var/log/messages) I see the following error that I can't explain. I tried archiving a directory just for the heck of it, but same message:
2017-08-18T09:54:16-07:00 isi01-1 isi_job_d[4784]: CloudPoolsTreewalk[8408] Waiting
2017-08-18T09:54:16-07:00 isi01-1 isi_job_d[4784]: CloudPoolsTreewalk[8408] Running
2017-08-18T09:54:16-07:00 isi01-1 isi_job_d[4784]: CloudPoolsTreewalk[8408] Phase 1: begin cloud treewalk
2017-08-18T09:54:16-07:00 isi01-1 isi_job_d[4784]: treewalk_assign_lins:472: [8408] Skipping lin 1:0d74:2aa3 isi01/cptest/pic1.jpg: not a directory
2017-08-18T09:54:16-07:00 isi01-1 isi_job_d[4784]: CloudPoolsTreewalk[8408] Phase 1: end cloud treewalk
2017-08-18T09:54:17-07:00 isi01-1 isi_job_d[4784]: CloudPoolsTreewalk[8408] Succeeded
Anyone else experience this? Shouldn't that path actually be /ifs/isi01/cptest/pic1.jpg or is it relative and this expected? I don't have anything to compare with since this is our first attempt with CloudPools. Thanks.
Phil.Lam
3 Apprentice
•
625 Posts
0
August 22nd, 2017 15:00
Ryan_CSULB,
Do you have a file policy for cloudpools? Checkout Isilon OneFS 8.0 CloudPools and ECS demo.
From https://www.emc.com/collateral/TechnicalDocument/docu65065.pdf:
The following example archives an entire directory to the cloud. The operation must match an existing file pool policy to be successful.
# isi cloud archive /ifs/data/images/*.* --recursive yes
sluetze
2 Intern
•
300 Posts
0
August 21st, 2017 05:00
2017-08-18T09:54:16-07:00 isi01-1 isi_job_d[4784]: treewalk_assign_lins:472: Skipping lin 1:0d74:2aa3 isi01/cptest/pic1.jpg: not a directory
Try it with a directory and not with a file?
Rgds
-- sluetze
oyo44
1 Rookie
•
8 Posts
0
August 21st, 2017 06:00
Are all of your nodes connected to the network? Or do you have NANON or NANOAN configuration?
Running Isilon OneFS in "Not All Nodes On Network" (NANON) and "Not All Nodes On All Networks" (NANOAN) configurations
Ryan_CSULB
1 Rookie
•
64 Posts
0
August 21st, 2017 09:00
sluetze, yes I tried doing the entire ctptest directory just to see what would happen, but got the same message about cptest not being a directory...
oyo44, all nodes definitely have access to the same network over 10GigE... I'm not isolating a given node and am not doing a split network at this time (using the gigabit connections). I'm not using LACP either, each node has 2 independent 10GigE connection paths.
My network has pretty strict firewall rules in place, and I've only opened up the smartconnect pool0 that I consider the mgmt pool (with access to SSH and WebGUI). I'm initiating the archive via one of the addresses on this pool0 as root (system access zone), I assume it doesn't care about any of the other access zones I have defined.
Ryan_CSULB
1 Rookie
•
64 Posts
0
August 23rd, 2017 14:00
Thanks Phil, that did it. I see now that I needed to define a VALID file pool policy for my files/directory before I could manually archive it. Your file must meet the criteria you set for the policy even to manually archive. I figured regardless of the policy if I tell it to archive, it would archive. Apparently not.
Interestingly enough, once I had a valid policy that "not a directory" error went away on archive (both firing the job and doing it manually). However, when I did a recall of the files, I got the "not a directory" message again in the logs, but the recall appears to have worked just fine. Not sure what that message is trying to say, but once things are set correctly, it seems not to matter much if it appears in the logs or doesn't.
Thanks again, saved me a call into support.
sluetze
2 Intern
•
300 Posts
0
August 24th, 2017 01:00
Could you file a SR via support anyway for the wrong Error-Coding / Error-Message?
Imho they should file a bug for this.
Ryan_CSULB
1 Rookie
•
64 Posts
0
August 24th, 2017 08:00
Yes, I can do that but before I do is there anywhere else I should check logs for this operation to see if a better error is thrown? I was looking in /var/log/messages. Since I don't get much there when the operation is a success (on archive) I wonder if it is logged somewhere else as well? Anyone know?
Phil.Lam
3 Apprentice
•
625 Posts
0
August 16th, 2018 14:00
Ryan_CSULB,
do a ls -lo on Isilon and if a files shows sstubbed, then it's archived.
Example:
# ls -lo
...
docu51121_MANAGING SMB SHARES AND HOME DIRECTORIES.pdf
-rw-r--r-- 1 root wheel inherit,writecache,wcinherit,sstubbed,hasads 9295 81 Oct 10 2017
docu84274_OneFS 8.1.0 CloudPools Administration Guide.pdf
-rw-r--r-- 1 root wheel inherit,writecache,wcinherit,sstubbed,hasads 4820 21 Oct 10 2017 .list_cloudpools_files_17-May-2018.zip
-rw-r--r-- 1 root wheel inherit,writecache,wcinherit,sstubbed,hasads 20920 47 Apr 20 19:59 virtustream.01 updated.pdf
...