-
Notifications
You must be signed in to change notification settings - Fork 1.2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
seastar - Exceptional future ignored: std::out_of_range (_Map_base::at) #1656
Comments
@asias I think I also have seen this in cases that nodes have data and we do a restart .... |
any other log messages, the issue looks strange ... |
@asias @slivne There's a full backtrace of this issue on the mailing list captured by @gleb-cloudius himself! |
On Thu, Sep 08, 2016 at 10:20:36AM -0700, Asias He wrote:
https://groups.google.com/forum/#!topic/scylladb-dev/YZjSRf8RU4s
|
I can't reproduce neither locally or on AWS. |
@slivne find a reproducer |
@gleb-cloudius - you managed to get this locally, no? What did you run? |
On Wed, Sep 14, 2016 at 02:51:30AM -0700, Calle Wilund wrote:
|
Ok, checking a little further: out_of_range("_Map_base::at") is thrown (not surprisingly) in std::__detail::_Map_base. However, this is only used as super type in _Hashtable derivatives, i.e. only std::unordered_map. Now, there is an unordered_map in the collectd loop, in the "cpwriter" that outputs data to packets. However, it does not use the "at" method. In fact, it only uses operator[] and "clear". So unless there is serious library bug or compiler weirdness going on, I don't see how that could be the culprit either. |
how to reproduce
I did it 5 times and on every restart I got the error. |
Pasting the snippet from scylladb-users groupsoftware: scylla-server-1.3.0-20160824.ec3ace5.el7.centos.x86_64 setup: 3 racks, 2 nodes per rack, 1 seed node per rack test: start the full 6 node cluster and then shutdown the 6 node cluster .. repeat..
In both cases above, I flush the nodes prior to shutting down the services. I'm not doing any IO between the start and stop. Just cluster up and cluster down. In the failure case, I see continuous stream of following messages on the nodes during startup (results in startup failure) Sep 18 11:33:16 XXX scylla[19483]: [shard 0] seastar - Exceptional future ignored: std::out_of_range (_Map_base::at) Again, above issue does not happen if I start the cluster without any data or commitlog directories i.e. clean. This is very easily recreatable and happens every single time. I'm not suspecting anything wrong with configuration files since the cluster comes up successfully when started without the data and commitlog dirs. Initially I did not have "nodetool flush" before the service shutdown so I added it but it didn't help. Are there are "startup ordering" constraints in the current version ? Any help is appreciated -krishna |
@asias, @ eyalgutkind https://github.com/eyalgutkind I am not sure the issue is I have created scylldb/scylla#1679 to follow this one On Tue, Sep 20, 2016 at 1:45 AM, Krishnanand Thommandra <
|
So, there are two issues:
|
Here is better trace: #0 0x00007ffff7251eb0 in std::out_of_range::out_of_range(char const_)@plt () from /lib64/libstdc++.so.6
|
Thanks gleb @AssiaS - collectd info related to gossip ? On Tue, Sep 20, 2016 at 3:41 PM, Gleb Natapov notifications@github.com
|
@gleb-cloudius Thanks. I think I found the culprit. |
Gleb, can you try this patch if you can reproduce locally?
|
It is possible that endpoint_state_map does not contain the entry for the node itself when collectd accesses it. Fixes the issue: Sep 18 11:33:16 XXX scylla[19483]: [shard 0] seastar - Exceptional future ignored: std::out_of_range (_Map_base::at) Fixes #1656 Message-Id: <8ffe22a542ff71e8c121b06ad62f94db54cc388f.1474377722.git.asias@scylladb.com> (cherry picked from commit aa47265)
On Tue, Sep 20, 2016 at 06:17:25AM -0700, Asias He wrote:
|
@gleb-cloudius thanks for confirming. |
…cross versions (2021.1)' from Eliran Sinvani This is the 2021.1 version of scylladb#1656 It has a lot less commits because almost all of the API commits and the reverted commits (from 2020.1) are already there. Closes scylladb#1660 * github.com:scylladb/scylla-enterprise: Merge 'Fix inconsistencies in MV and SI (reworked)' from Eliran Sinvani storage_proxy: Add .local_db() getters
Installation details
Scylla version (or git commit hash):
Cluster size: 3
OS (RHEL/CentOS/Ubuntu/AWS AMI): ScyllaDB 1.3 (ami-78d7b46f)
Instance c3.large
on initial boot we get:
I was not able to reproduce this with ccm easily.
The text was updated successfully, but these errors were encountered: