deployment_head stuck after reopen allocation

[SpontaneousOverthrow]

After reopen allocation deployment_head metric didn't update value(stuck on latest value before close allocation)
In db subgraph is healthy and latest_ethereum_block_number is equal to ethereum_chain_head_number
deployment,latest_ethereum_block_number,last_healthy_ethereum_block_number,health,synced from subgraphs.subgraph_deployment; deployment | latest_ethereum_block_number | last_healthy_ethereum_block_number | health | synced ------------------------------------------------+------------------------------+------------------------------------+---------+-------- QmUghXvKf5cVjtayNNRHCd3RbHEwfbGBQ95s9vheJjN5hH | 12780253 | | healthy | t
This happened for multi subraphs on each reopen.
Index node restart helps.
Using graph-node:v0.23.1

[cryptovestor21]

Confirming that I also see the same behaviour, and restarting the index nodes seems to resolve the issue.

[azf20]

Thanks @SpontaneousOverthrow and @cryptovestor21 . @fordN is there any reason that allocations should impact the deployment head metric in graph-node?

[chriswessels]

Confirming I am seeing the same behaviour. Not sure if it was reallocations that broke deployment_head. In theory, a reallocation may cause indexer-agent to ensure the deployment is made, and maybe those interactions with the graph-node API are causing the metric to break? Happy to help debug

[begetan]

I confirm this issue regarding the graph-node monitoring endpoint.
The only difference is that restart doesn't help, so I've created a separate issue: #2699