1. Judging from the situation described, initial sync encounters performance problems; STARTUP2 indicates that the node is doing initial sync;
2. Therefore, you can consider optimizing initial sync; if you have not done so, you can consider backing up/restoring the database to the node to be added, and then adding the node to reduce the sync data;
3. How many ways are there for backup/recovery? You can choose a familiar method; Ops Manager is actually a good choice.
I don’t know what the machine configuration is. Are you sure it was killed by the system’s OOM killer? If it is an oom killer, you can add swap to avoid this problem. In addition, @bguo has mentioned that you can try to restore a node from backup. Please refer to the documentation for specific methods. But even so, a certain amount of swap is still recommended.
You should first use tools such as devtool to locate it. Profiles->Take Heap SnapshotIf a large number of characters are processed, you should consider whether there are strings that have not been released because they were placed in the upper-level scope
1. Judging from the situation described, initial sync encounters performance problems; STARTUP2 indicates that the node is doing initial sync;
2. Therefore, you can consider optimizing initial sync; if you have not done so, you can consider backing up/restoring the database to the node to be added, and then adding the node to reduce the sync data;
3. How many ways are there for backup/recovery? You can choose a familiar method; Ops Manager is actually a good choice.
For reference.
Love MongoDB! Have Fun!
I don’t know what the machine configuration is. Are you sure it was killed by the system’s OOM killer? If it is an oom killer, you can add swap to avoid this problem.
In addition, @bguo has mentioned that you can try to restore a node from backup. Please refer to the documentation for specific methods. But even so, a certain amount of swap is still recommended.
You should first use tools such as devtool to locate it.
Profiles->Take Heap Snapshot
If a large number of characters are processed, you should consider whether there are strings that have not been released because they were placed in the upper-level scope