[Openstack] Error uploading large objects(>1G) on swift

Shashank Sahni shredder12 at gmail.com
Fri Mar 8 11:36:42 UTC 2013


Hi,

Here  is a detailed overview of my setup.

Two machines(each 4G RAM and one 2TB disk)

   1. Configured as proxy + storage
   2. Configured as storage

I followed the folsom object storage administrator guide and created ring
partitions using the following commands.

# swift-ring-builder account.builder 9 2 1

# swift-ring-builder container.builder 9 2 1

# swift-ring-builder object.builder 9 2 1

I added both the storage units in the same zone, copied the *.ring.gz files
to each host and restarted the services.

The setup seems to be working fine for small files, but while copying large
files I get errors(as stated in the previous mail). Any help would be
greatly appreciated.

Thanks.

--
Shashank Sahni


On Fri, Mar 8, 2013 at 12:51 PM, Shashank Sahni <shredder12 at gmail.com>wrote:

> Hi everyone,
>
> I'm running a small test setup of swift with one proxy+storage and one
> storage instance.
>
> The setup seems to be running well for small objects. I'm able to create
> and upload them successfully using swift cli tool, but for large
> objects(>1G), I get error of "503 service unavailable"(on swift cli) after
> a long wait.
>
> I dig into logs and found that none of the replicator services are
> running. All of them give the following error(array index out of range)
>
> Mar  8 12:42:03 yahoo-128 account-replicator Beginning replication run
> Mar  8 12:42:03 yahoo-128 account-replicator STDOUT: Traceback (most
> recent call last):
> Mar  8 12:42:03 yahoo-128 account-replicator STDOUT: File
> "/usr/lib/python2.7/dist-packages/eventlet/greenpool.py", line 80, in
> _spawn_n_impl
> Mar  8 12:42:03 yahoo-128 account-replicator STDOUT: func(*args, **kwargs)
> Mar  8 12:42:03 yahoo-128 account-replicator STDOUT: File
> "/usr/lib/python2.7/dist-packages/swift/common/db_replicator.py", line 398,
> in _replicate_object
> Mar  8 12:42:03 yahoo-128 account-replicator STDOUT: nodes =
> self.ring.get_part_nodes(int(partition))
> Mar  8 12:42:03 yahoo-128 account-replicator STDOUT: File
> "/usr/lib/python2.7/dist-packages/swift/common/ring/ring.py", line 201, in
> get_part_nodes
> Mar  8 12:42:03 yahoo-128 account-replicator STDOUT: if not (r[part] in
> seen_ids or seen_ids.add(r[part]))]
> Mar  8 12:42:03 yahoo-128 account-replicator STDOUT: IndexError: array
> index out of range
> Mar  8 12:42:03 yahoo-128 account-replicator STDOUT: Traceback (most
> recent call last):
> Mar  8 12:42:03 yahoo-128 account-replicator STDOUT: File
> "/usr/lib/python2.7/dist-packages/eventlet/greenpool.py", line 80, in
> _spawn_n_impl
> Mar  8 12:42:03 yahoo-128 account-replicator STDOUT: func(*args, **kwargs)
> Mar  8 12:42:03 yahoo-128 account-replicator STDOUT: File
> "/usr/lib/python2.7/dist-packages/swift/common/db_replicator.py", line 398,
> in _replicate_object
> Mar  8 12:42:03 yahoo-128 account-replicator STDOUT: nodes =
> self.ring.get_part_nodes(int(partition))
> Mar  8 12:42:03 yahoo-128 account-replicator STDOUT: File
> "/usr/lib/python2.7/dist-packages/swift/common/ring/ring.py", line 201, in
> get_part_nodes
> Mar  8 12:42:03 yahoo-128 account-replicator STDOUT: if not (r[part] in
> seen_ids or seen_ids.add(r[part]))]
> Mar  8 12:42:03 yahoo-128 account-replicator STDOUT: IndexError: array
> index out of range
> Mar  8 12:42:03 yahoo-128 account-replicator Replication run OVER
> Mar  8 12:42:03 yahoo-128 account-replicator Attempted to replicate 3 dbs
> in 0.02536 seconds (118.26332/s)
> Mar  8 12:42:03 yahoo-128 account-replicator Removed 0 dbs
> Mar  8 12:42:03 yahoo-128 account-replicator 1 successes, 0 failures
> Mar  8 12:42:03 yahoo-128 account-replicator no_change:1 ts_repl:0 diff:0
> rsync:0 diff_capped:0 hashmatch:0 empty:0
>
> Additionally, while the transfer of a big file was going on(4GB), I got
> the following error in logs.
>
> proxy-server ERROR with Object server 10.0.1.1:6000/sdb1 re: Trying to
> write to
> /v1/AUTH_cccc28c949f34bd3b5fb295cffb69ed4/data_dir/data_dir/Feb27/PRIVATE/AVCHD/BDMV/STREAM/00002.MTS:
> ChunkWriteTimeout (10s)
>
> Any suggestions?
>
> --
> Shashank Sahni
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20130308/ea791f66/attachment.html>


More information about the Openstack mailing list