Restoring a file dies with this message

[tk422]

Hey again,

Went to test the restore of a tar.gz file,now using a long name without the double tar.gz. Can you actually add a period back after the last part of the filename that way the date code doesnt get jumbled together.

Here is what I get:

2013-02-26 16:33:13,182 INFO: Restoring gridlock.XXXX.org.23420130221100557.tgz.enc <*_**-- note the 234 is the last part of the file name and would like that separate from the date
Password:
2013-02-26 16:33:43,045 INFO: Downloading...
Traceback (most recent call last):
File "/usr/local/bin/bakthat", line 9, in
load_entry_point('bakthat==0.3.7', 'console_scripts', 'bakthat')()
File "/usr/local/bin/bakthat.py", line 598, in main
app.run()
File "/usr/local/lib/python2.6/dist-packages/aaargh.py", line 157, in run
return func(__kwargs)
File "/usr/local/bin/bakthat.py", line 506, in restore
out = storage_backend.download(key_name, *_download_kwargs)
File "/usr/local/bin/bakthat.py", line 267, in download
job = self.vault.retrieve_archive(archive_id)
File "/usr/local/lib/python2.6/dist-packages/boto/glacier/vault.py", line 293, in retrieve_archive
response = self.layer1.initiate_job(self.name, job_data)
File "/usr/local/lib/python2.6/dist-packages/boto/glacier/layer1.py", line 351, in initiate_job
response_headers=response_headers)
File "/usr/local/lib/python2.6/dist-packages/boto/glacier/layer1.py", line 84, in make_request
raise UnexpectedHTTPResponseError(ok_responses, response)
boto.glacier.exceptions.UnexpectedHTTPResponseError: Expected 202, got (400, code=InvalidParameterValueException, message=Invalid vault name: arn:aws:glacier:us-east-1:917509789911:vaults/backups-east)

[tk422]

looking into this further it looks like calling via a python module does not honor the glacier vault (hence the error about finding the file). Here is what I have in the upload script and (I just checked none of them are uploaded to the vault but instead are uploaded to "None")

vaultname = "backups-east"

aws_conf = {"access_key":ACCESS_KEY_ID, "secret_key":SECRET_ACCESS_KEY, "bucket": "", "vault": vaultname, "password":encpassword}

for each in os.listdir(bakfolder):
for files in os.listdir(os.path.join(bakfolder,each)):
if files[-3:] != ".gz":
pass
else:
bakthat.backup(os.path.join(bakfolder,each,files), conf=aws_conf, destination="glacier", password=encpassword, prompt="no")

[tsileo]

First, I totally agree with the dot before the date, I'll make the change quickly,

About your fist issue, the error says: Invalid vault name, I think you shouldn't have a dash in your vault name (backups-east), it's a boto/AWS related issue I think.

Now about the bakthat module call, I use bakthat in Python scripts to make daily backups to glacier on few servers, and I never had any issue:

• first, the prompt argument is useless when calling bakthat from Python (either password="" to disable password, either a valid password)
• next, if you already have a configuration file, you don't have to pass the conf argument, and when passing it, I suggest you to set a S3 bucket just to backup your local inventory glacier (if something goes wrong you don't have to wait 4-5 hours just to retrieve your inventory.)

Can you try without the conf (if you have a ~/.bakthat.conf) and promp argument, and show me the logs output, to check bakthat logs when using a Python script, just add these line at the top of your file:

import logging
log = logging.getLogger()
log.setLevel(logging.INFO)

Also, I'll rewrite the doc for both command line and python module usage, with backups scripts example.

Keep me updated, thanks for your feedback !

[tsileo]

I made the change on github to add a dot betwen the backup name and the date component, I have not pushed the change on pypi yet.

[tk422]

Okay let me play around with it later today. I know that when I backup and restore from the command line (taking the configuration from .bakthat.cfg file) it works fine with the dash in the glacier name. But let me try a couple of things and see if I can pinpoint the breakdown some more.

[tk422]

Okay when I call it via python module as shown above it doesnt work it errors out with the same error seen above. But when I do it via the commandline it correctly uploads it to the vault backups-east. So there might be something with the dash in the name when called via the module. Thats my current guess anyway. I'll look more into it.

[tk422]

Tested it a few more times. It does look like its not properly uploading to a vault with a dash in it when using the module. But seems to work on the commandline.

[tk422]

Any thoughts on this? I have been moving so havent been able to look at the code.

[tsileo]

It seems to be a boto related issue, so I will investigate more when bakthat 0.4.0 will be released and post an issue on boto repository if needed, I will keep you updated.

I making big changes on bakthat right now (not updated on pypi yet) with version 0.4.0 release coming soon and I would love to hear your feedback.

[vsespb]

FYI Similar issue vsespb/mt-aws-glacier#32 https://forums.aws.amazon.com/thread.jspa?threadID=123076&tstart=0

Also I was unable to receive error message "Invalid vault name" in the wild - message is not thrown when vault name is really not valid.

Instead:

1. When not found:

{"code":"ResourceNotFoundException","message":"Vault not found for ARN: arn:aws:glacier:us-east-1:XXXXXXXX:vaults/test1zz","type":"Client"}

2. When name contains invalid characters:

{"code":"ResourceNotFoundException","message":"Vault not found for ARN: testz:z","type":"Client"}

3. When permission problem: AccessDeniedException

[vsespb]

FYI it turns out that Invalid vault name means that this archive_id belongs to different vault https://forums.aws.amazon.com/thread.jspa?threadID=123076&tstart=0

[tsileo]

Thanks for the follow up, I'll try to update bakthat accordingly !