docs: Clarify H2 database caching strategy

[chadlwilson]

Description of Change

• Fixes a few typos for repository for RetireJS
• Attempts to clarify the intent of the first H2 DB caching strategy
  • I understand this is about caching JUST the H2 DB alone, not the entire data directory
  • Theoretically, I think you could use the single node approach caching either the H2 DB alone OR the entire data directory, so it's possible there needs to be some rewording or re-titling of the strategies here
  • It's possible there are actually three strategies here?
    1. Caching H2 DB alone with single node update
    2. Caching entire data dir with single node update
    3. Caching entire data dir with multiple/all node update (all nodes can upload archive)

Is the intent of the two strategies to primarily focus on

1. single node vs multiple node update, OR
2. to focus on whether the H2 DB vs. the entire data dir including cache is archived (and other files such as publishedSuppressions.xml, jsrepository.json etc)

?

I'll help reword/adjust to meet your intention :-)

[jeremylong]

LGTM

[jeremylong]

I was trying to focus on single node update vs multiple node updating. If you scan with autoupdate off and you don't cache the retire js JSON file, etc. there could be problems.

[chadlwilson]

OK, in that case do we want to specifically recommend that users are best to archive the entire data directory (including cache and oss_cache subfolders) in both single node and multi-node strategies (to keep it simple for them)?

Or are there circumstances in which you would not want to include the cache directories (perhaps when they become "build specific" and might actually break ODC in some way?) so we can possibly give users some guidance as to how to make the decision on what to include in the archive.

Zipping an entire directory is obviously easier than filtering it, unless we can give more specific guidance on what you might need to filter without constraining the future development too much by creating an implicit "contract" on how the data dir is used and structured.

[jeremylong]

likely just zipping the entire data directory is best. If you use a single update node - there won't be anything in the cache if the only thing it is doing is updating the db...

[chadlwilson]

OK, thanks. How does that now read to you? Hope I have covered your intent.

[jeremylong]

LGTM