usage: rucio-necromancer [-h] [--run-once] [--threads THREADS] [--bulk BULK]
[--sleep-time SLEEP_TIME]
The Necromancer daemon is responsible for managing bad replicas. If a replica that got declared bad has other replicas, it will try to recover it by requesting a new transfer. If there are no replicas anymore, then the file gets marked as lost.
optional arguments:
-h, --help show this help message and exit
--run-once Runs one loop iteration
--threads THREADS Concurrency control: number of threads
--bulk BULK Bulk control: number of requests per cycle
--sleep-time SLEEP_TIME
Concurrency control: thread sleep time after each
chunk of work
Lost replica:
In this example the file gets uploaded and will only have this replica as there are no replication rules. If it gets declared bad, there will be no replica to recover from.
Therefore the replica gets marked as lost.
Upload a file::
$ rucio upload --scope mock --rse MOCK --name file filename.txt
Check replicas::
$ rucio list-file-replicas mock:file
+---------+--------+------------+-----------+---------------------------------------------------------+
| SCOPE | NAME | FILESIZE | ADLER32 | RSE: REPLICA |
|---------+--------+------------+-----------+---------------------------------------------------------|
| mock | file | 149.000 B | 948240 | MOCK: file://localhost:0/tmp/rucio_rse/mock/fb/d1/file |
+---------+--------+------------+-----------+---------------------------------------------------------+
Declare it as bad::
$ rucio-admin replicas declare-bad file://localhost:0/tmp/rucio_rse/mock/fb/d1/file --reason 'bad'
Run the daemon::
$ rucio-necromancer --run-once
Check replicas again::
$ rucio list-file-replicas mock:file
+---------+--------+------------+-----------+----------------+
| SCOPE | NAME | FILESIZE | ADLER32 | RSE: REPLICA |
|---------+--------+------------+-----------+----------------|
+---------+--------+------------+-----------+----------------+
Bad replica:
In this example the file gets uploaded and will have two replicas. If it gets declared bad, then the daemon will try to recover it from the second replica.
Upload a file and replicate it::
$ rucio upload --scope mock --rse MOCK filename.txt
$ rucio add-rule mock:file 1 MOCK2
$ rucio-conveyor-submitter --run-once
Check replicas::
$ rucio list-file-replicas mock:file
+---------+--------+------------+-----------+---------------------------------------------------------+
| SCOPE | NAME | FILESIZE | ADLER32 | RSE: REPLICA |
|---------+--------+------------+-----------+---------------------------------------------------------|
| mock | file | 149.000 B | 948240 | MOCK: file://localhost:0/tmp/rucio_rse/mock/fb/d1/file |
|---------+--------+------------+-----------+---------------------------------------------------------|
| mock | file | 149.000 B | 948240 | MOCK2: file://localhost:1/tmp/rucio_rse/mock/fb/d1/file |
+---------+--------+------------+-----------+---------------------------------------------------------+
Declare one replica as bad::
$ rucio-admin replicas declare-bad file://localhost:1/tmp/rucio_rse/mock/fb/d1/file --reason 'bad'
Run the daemon::
$ rucio-necromancer --run-once
Check replicas again::
$ rucio list-file-replicas mock:file
+---------+--------+------------+-----------+---------------------------------------------------------+
| SCOPE | NAME | FILESIZE | ADLER32 | RSE: REPLICA |
|---------+--------+------------+-----------+---------------------------------------------------------|
| mock | file | 149.000 B | 948240 | MOCK: file://localhost:0/tmp/rucio_rse/mock/fb/d1/file |
|---------+--------+------------+-----------+---------------------------------------------------------|
| mock | file | 149.000 B | 948240 | MOCK2: file://localhost:1/tmp/rucio_rse/mock/fb/d1/file |
+---------+--------+------------+-----------+---------------------------------------------------------+