Description of problem:
The migration of OpenShift data from etcd2 -> etcd3 currently fails while running the etcdctl migrate tool. It errors out immediately whether a custom transformer or the default one is used.
The error (with some extra debugging in the custom transformer used) is:
Starting...
ERROR CODING NODE 0 {Key: , CreatedIndex: 0, ModifiedIndex: 0, TTL: 0}
Reader(0)[write /dev/stdin: bad file descriptor] =
Err: write /dev/stdin: bad file descriptor
Error: exit status 1
failed to transform keys
Version-Release number of selected component (if applicable): 3.4.0.11 and etcd 2.3.7 -> etcd3 3.0.12
How reproducible: always (at least on the OCP 3.4 data sets we have tried)
Steps to Reproduce:
1. Install OCP 3.4.0.11 with a separate etcd 2.3.7
2. Create some projects, pods, etc. General OCP data
3. Backup the etcd environment
4. yum swap etcd etcd3 and verify the OCP env is still operational running etcd3 in etcd2 mode
5. etcdctl --endpoints $endpoints migrate --data-dir=/var/lib/etcd
Actual results: Error message (as above)
Expected results:
Successful migration
Comment 3Timothy St. Clair
2016-10-20 14:19:07 UTC
fix is here: https://github.com/coreos/etcd/pull/6690, we'll need to update our etcdctl binaries.
Hold off on final update until we've had a bit more time to validate.
Hi Mike,
Can you help to provide some data for etcd for testing this? Seems I can not find OCP 3.4 env for testing it. And I try to reproduce it with some key value with '%' such as "test%" and "%", then migrate datedir from etcd-2.3.7-4 to etcd3-3.0.12-3, also borrow a openshift 3.3 env to test it. But it won't trigger the problem.
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.
https://rhn.redhat.com/errata/RHBA-2016-2890.html
Description of problem: The migration of OpenShift data from etcd2 -> etcd3 currently fails while running the etcdctl migrate tool. It errors out immediately whether a custom transformer or the default one is used. The error (with some extra debugging in the custom transformer used) is: Starting... ERROR CODING NODE 0 {Key: , CreatedIndex: 0, ModifiedIndex: 0, TTL: 0} Reader(0)[write /dev/stdin: bad file descriptor] = Err: write /dev/stdin: bad file descriptor Error: exit status 1 failed to transform keys Version-Release number of selected component (if applicable): 3.4.0.11 and etcd 2.3.7 -> etcd3 3.0.12 How reproducible: always (at least on the OCP 3.4 data sets we have tried) Steps to Reproduce: 1. Install OCP 3.4.0.11 with a separate etcd 2.3.7 2. Create some projects, pods, etc. General OCP data 3. Backup the etcd environment 4. yum swap etcd etcd3 and verify the OCP env is still operational running etcd3 in etcd2 mode 5. etcdctl --endpoints $endpoints migrate --data-dir=/var/lib/etcd Actual results: Error message (as above) Expected results: Successful migration