On Monday, 21 May 2018 01:56:18 UTC+2, Patrick McShane  wrote:
> Hi,
> 
> 
> We've setup an Amazon S3 bucket with nearly wide open permissions.
> 
> 
> We supplied all the parameters as per documentation.
> 
> 
> When we start a test backup job it seems to queue and then enter a "running" 
> state for a long period of time with no obvious failures or error 
> codes/messages.
> 
> 
> After the job has been running for a few minutes, a subsequent check of 
> messages from the bconsole CLI shows that a mount request exists for the S3 
> volume because the mount was not automatic.  We then honor the S3 volume 
> mount request using console "mount" command only to see another mount 
> request. 
> 
> 
> Has anyone experienced this "mount request loop" when testing the 
> bareos-storage-droplet S3 addon package?
> 
> 
> Any general recommendations on what configuration options are causing this 
> repeating mount request behavior when attempting to write to an S3 object 
> store? 
> 
> 
> Any recommendations on AWS S3 bucket attributes on the Amazon side that may 
> fix this? 
> 
> 
> Thanks, 
> Pat 

I have exactly the same issue.
I have turned on verbose monitoring on my S3 bucket which I am using for 
droplet and it is not even being hit by Bareos at all.
When I try the bucket with awscli or with S3FS, it works without any issues 
with exactly the same credentials, but it doesn't from S3 droplet.
AWS doesn't report any get/put/head requests, it doesn't return any 5xx nor 4xx 
errors.

I also tried adding "https://"; in the bucket address, but then it returns:
17-May-2018 13:36:00.402878 17-May-2018 13:36:00.402918 bareos-sd (850): 
message.c:1129-3865 DIRECTOR for following msg: bareos-sd: ERROR in 
droplet_device.c:103 error: src/addrlist.c:638: dpl_addrlist_add: cannot lookup 
host https: Unknown host
17-May-2018 13:36:00.402979 bareos-storage-lon-sd (10): 
droplet_device.c:103-3865 ERROR: error: src/profile.c:293: conf_cb_func: error 
parsing address list
17-May-2018 13:36:00.403006 bareos-sd (850): message.c:858-3865 Enter 
dispatch_message type=4 msg=bareos-storage-lon-sd: ERROR in 
droplet_device.c:103 error: src/profile.c:293: conf_cb_func: error parsing 
address list
17-May-2018 13:36:00.403036 bareos-sd (850): message.c:1129-3865 DIRECTOR for 
following msg: bareos-sd: ERROR in droplet_device.c:103 error: 
src/profile.c:293: conf_cb_func: error parsing address list
17-May-2018 13:36:00.403096 bareos-sd (100): droplet_device.c:704-3865 
droplet_device.c:703 Failed to create a new context using config 
profile=/etc/bareos/bareos-sd.d/device/droplet/droplet.profile,bucket=name-of-my-bucket,iothreads=1,ioslots=1,chunksize=100M

I have tried switching use_https to False too, but it didn't change anything, 
exactly the mount loop was happening.
This is what setdebug=999 for storage reports when the loop happens:
16-May-2018 09:55:49.133920 bareos-sd (100): droplet_device.c:292-0 Flushing 
chunk /S3DropletTest-Full-2886/0001
16-May-2018 09:55:49.158134 bareos-sd (100): chunked_device.c:413-0 Enqueueing 
chunk 1 of volume S3DropletTest-Full-2886 for retry of upload later
16-May-2018 09:55:49.166766 bareos-sd (100): chunked_device.c:413-0 Enqueueing 
chunk 0 of volume S3DropletTest-Full-2886 for retry of upload later
16-May-2018 09:59:15.203428 bareos-sd (100): chunked_device.c:398-0 Flushing 
chunk 0 of volume S3DropletTest-Full-2906 by thread 0x6100007f86617fa7
16-May-2018 09:59:15.203475 bareos-sd (100): droplet_device.c:292-0 Flushing 
chunk /S3DropletTest-Full-2906/0000
16-May-2018 09:59:15.231959 bareos-sd (100): chunked_device.c:398-0 Flushing 
chunk 1 of volume S3DropletTest-Full-2906 by thread 0x3900007f8660ff97
16-May-2018 09:59:15.231995 bareos-sd (100): droplet_device.c:292-0 Flushing 
chunk /S3DropletTest-Full-2906/0001
16-May-2018 09:59:15.271106 bareos-sd (100): chunked_device.c:413-0 Enqueueing 
chunk 0 of volume S3DropletTest-Full-2906 for retry of upload later
16-May-2018 09:59:15.306617 bareos-sd (100): chunked_device.c:413-0 Enqueueing 
chunk 1 of volume S3DropletTest-Full-2906 for retry of upload later

First I had used 5 threads with 5 slots and for 3 jobs in parallel, so I 
thought I have reached S3 bucket limits for requests, but then I changed to 1 
thread with 1 slot and the issue remained exactly the same - this was the 
moment when I started to check AWS monitoring for buckets.

I'd love to move on with that.
If needed, I can provide obfuscated configs related to the droplet backend.

Best Regards,
Jan.

-- 
You received this message because you are subscribed to the Google Groups 
"bareos-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to