# Warm nodes on spinning disks not stable, falling out of cluster during recovery

**URL:** <https://discuss.elastic.co/t/warm-nodes-on-spinning-disks-not-stable-falling-out-of-cluster-during-recovery/250078>\
**Category:** Elasticsearch\
**Created:** [September 27, 2020, 11:34am UTC](https://discuss.elastic.co/t/warm-nodes-on-spinning-disks-not-stable-falling-out-of-cluster-during-recovery/250078 "2020-09-27T11:34:03Z")\
**Posts on this page:** 1\
**Showing post:** 4

<div class="post-metadata">

**Author:** ![DavidTurner](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/davidturner/32/22453_2.png) [@DavidTurner](https://discuss.elastic.co/u/DavidTurner)\
**Post date:** [September 28, 2020, 6:36am UTC](https://discuss.elastic.co/t/warm-nodes-on-spinning-disks-not-stable-falling-out-of-cluster-during-recovery/250078/4 "2020-09-28T06:36:38Z")

</div>

It looks like you are pushing your nodes far harder than the defaults allow, and far harder than they can cope:

> [@otoce](#):
>
> ```auto
> "cluster.routing.allocation.node_concurrent_incoming_recoveries" : "4",
> 
> ```

The default for this is `2` and increasing it is not recommended.

> [@otoce](#):
>
> ```auto
> "cluster.routing.allocation.node_concurrent_outgoing_recoveries" : "4",
> 
> ```

The default for this is `2` and increasing it is not recommended.

> [@otoce](#):
>
> ```auto
> "cluster.routing.allocation.node_concurrent_recoveries" : "8",
> 
> ```

The default for this is `2` and increasing it is not recommended.

> [@otoce](#):
>
> ```auto
> "cluster.routing.allocation.node_initial_primaries_recoveries" : "40",
> 
> ```

The default for this is `4` and increasing it is not recommended.

> [@otoce](#):
>
> ```auto
> "indices.recovery.max_bytes_per_sec" : "250mb"
> 
> ```

The default for this is `40mb`; increasing it is ok if your nodes can cope with the increase, but the evidence indicates that `250mb` is far too high for your warm nodes.

---

_[View the full topic](https://discuss.elastic.co/t/warm-nodes-on-spinning-disks-not-stable-falling-out-of-cluster-during-recovery/250078)._
