docs: extend waku store config tutorial of the operator docs #1359
Conversation
LNSD
added
documentation
Jenkins BuildsClick to see older builds (10)
|
LNSD
force-pushed
the
docs-store-config
branch
from
3f5cc52 to
53a4213
Compare
|
|
||
| ## Configuring the node as a store service node | ||
|
|
||
| If the waku store node is enabled (the `--store` option is set to `true`) the node will store historical messages and will be able to serve those messages to the waku store clients. |
There was a problem hiding this comment.
Does that mean storing messages is default? (Because --store=true is default)
There was a problem hiding this comment.
I'd make it explicit above that you need to set --store=false and set --storenode to setup a client only store-node.
There was a problem hiding this comment.
Does that mean storing messages is default? (Because
--store=trueis default)
Yes, that's what I meant with "the node will store historical messages".
I'd make it explicit above that you need to set
--store=falseand set--storenodeto setup a client only store-node.
This is not the case. Both things (server and client) are independent. You can have both on, both off, or one on and the other off.
This is what I meant with this: https://github.com/status-im/nwaku/pull/1359/files?diff=split&w=0#diff-a2d58beafcccb9e2b626c20cb0e31383f1107c520e7a427e20d2896763f71eb7R16 😁
There was a problem hiding this comment.
True, but it may be worth spelling out in the client section that to configure a client only store must explicitly be set to false. This is clearly implied, of course, but we should assume that operators will just skip to the section they're interested in and may be surprised that following the client guidelines lead to a node that's also persisting messages.
Not part of this conversation, but wondering if shouldn't rather make the default for store false. Otherwise many operators upgrading after the release will suddenly start persisting messages. We can warn them, of course, but I think the sensible default for a nwaku node should just be a relay node.
There was a problem hiding this comment.
LGTM! Thanks. Apologies for only reviewing after merging. Comments below are not critical at all and can be addressed in future. A good idea to follow semantic line breaks in future. :)
| Currently store service nodes use, by default, a message store backed by an in-disk SQLite database. Most Waku messages average a size of 1KB - 2KB, implying a minimum memory requirement of at least ~250MB |
There was a problem hiding this comment.
Great to have some numbers here!
| * `--store-message-retention-policy`: This option controls the retention policy i.e., how long certain messages will be persisted. Two different retention policies are supported: | ||
| + The time retention policy,`time:<duration-in-seconds>` (e.g., `time:14400`) | ||
| + The capacity retention policy,`capacity:<messages-count>` (e.g, `capacity:25000`) | ||
| + To disable the retention policy, explicitly, set this option to to `""`, an empty string. |
There was a problem hiding this comment.
Can we spell out the implications? Something like: Disabling the retention policy implies that, from nwaku's perspective, messages will either be persisted indefinitely or that retention is managed by a third party outside of the node.
|
|
||
| ## Configuring the node as a store service node | ||
|
|
||
| If the waku store node is enabled (the `--store` option is set to `true`) the node will store historical messages and will be able to serve those messages to the waku store clients. |
There was a problem hiding this comment.
True, but it may be worth spelling out in the client section that to configure a client only store must explicitly be set to false. This is clearly implied, of course, but we should assume that operators will just skip to the section they're interested in and may be surprised that following the client guidelines lead to a node that's also persisting messages.
Not part of this conversation, but wondering if shouldn't rather make the default for store false. Otherwise many operators upgrading after the release will suddenly start persisting messages. We can warn them, of course, but I think the sensible default for a nwaku node should just be a relay node.
This PR covers the work pending of #1293
Same here: