amtool: Add silence display subcommand, add silence-ids validation, fix nil pointer dereference.

[yurkeen]

This change brings missing amtool functionality, which allows to request and display silence information by user-specified silence ID. Silence IDs are verified to be valid UUIDs, and when they are, retrieved from alertmanager. Before this change a user would need to filter silences by silence ID using other tools (think grep, jq, etc).

Naming for the subcommand, i.e. display, was chosen over show to not bring confusion to users, given that amtool already has query. Happy to change if any objections.

This PR also adds validation for silence-ids supplied to the silence display, silence update and silence expire subcommands, so when supplied values are not valid UUIDs, error is returned.

Additionally, this PR fixes a nil pointer dereference bug which happens due to the change in #2471, where the IsEqual *bool field of the Matcher model is optional and at the same time is intended to have true as its default value. Therefore, when a Matcher – with its IsEqual field set to nil– is received by the format/labelsMatcher() func, amtool panics.

Edit: typos

[yurkeen]

@simonpasquier or @w0rm – may I seek your reviews, when possible? Thanks!

[angelinelee22]

Hi,
I've been running into panic: runtime error: invalid memory address or nil pointer dereference whenever I run amtool silence query with either simple or extended formatting. Wanted to confirm if this is related to the above, although it runs fine with the -q quiet tag or -o json otherwise. If so, the issue has rendered amtool silence query unusable. It'd be great if this could be pushed along. Thanks!

[yurkeen]

Hi @angelinelee22,

It is very likely that you are facing the bug covered by this PR, if your alertmanager version is v0.22.0 or older, where #2602 was not yet fixed.

[angelinelee22]

Hi @yurkeen,
Sorry for missing this. I am actually using 0.22.2 on the darwin/amd64 platform and can confirm I'm still facing this issue.

[yurkeen]

Nill pointer dereference fixed in #2668. Since the rest did not get any attention, I'm closing the PR.