[skmonitor] Add monitoring API #201
Conversation
skiplabsdaniel
force-pushed
the
monitoring
branch
3 times, most recently
from
274f378
to
4c779b9
Compare
|
Bear with me here, I do plan to look at this. |
There was a problem hiding this comment.
This is great. I really like it.
Big +1 for using open telemetry - the data model (tree of spans, etc) looks solid to me, and we'll be able to utilise all the tools of that ecosystem.
Also, thanks for capturing all the memory info. That will be priceless.
What are your thoughts on the client? Due to the nature of the product, a lot of action is happening there. Should we log at all? Should we just expose to the user or should we try and capture this info somehow? Maybe make it opt in or out?
| } | ||
|
|
||
| fun read_line(): ?String { | ||
| .read_line().map(line -> { |
There was a problem hiding this comment.
Will this end up copying?
There was a problem hiding this comment.
I don't think so.
| module SKCSV; | ||
|
|
||
| fun read_line(): ?String { | ||
| SKDB.read_line() | ||
| } | ||
|
|
||
| fun skipExit<T>(res: Int): T { | ||
| SKDB.skipExit(res) | ||
| } | ||
|
|
||
| fun print_error(error: String): void { | ||
| SKDB.print_error(error) | ||
| } | ||
|
|
||
| fun print_raw(str: String): void { | ||
| SKDB.print_raw(str) | ||
| } | ||
|
|
||
| fun print_string<T: readonly Show>(value: T): void { | ||
| SKDB.print_string(value) | ||
| } |
There was a problem hiding this comment.
Did you intend to capture stuff here?
There was a problem hiding this comment.
We need to manage the skipExit for clean exit and readline to get the number of read bytes.
The rest is to be belt and braces.
skmonitor/src/Monitor.sk
Outdated
| if (!(monitoring() is MOn _)) return void; | ||
| SKStore.switchToTrace(); | ||
| low = get_session_low(); | ||
| high = get_session_high(); | ||
| traceId = TraceId::fromHighAndLow(high, low); | ||
| optParent = get_span() match { | ||
| | Some(parent) -> | ||
| saved = SKStore.newObstack(); | ||
| Some((parent, saved)) | ||
| | _ -> | ||
| initWriter(); | ||
| None() | ||
| }; | ||
| span = mutable Span( | ||
| now(), | ||
| traceId, | ||
| SpanId::fromInt(Random.next()), | ||
| optParent, | ||
| ); | ||
| set_span(Some(span)); | ||
| SKStore.restoreFromTrace(); |
There was a problem hiding this comment.
Should we put restore in a finally block? Same for elsewhere obviously.
There was a problem hiding this comment.
I don't think so, there should never have exceptions during trace calls.
|
And just one other thing: we don't have any logging in the production side. So on the server it's SqlTailer.sk and on the client it's the combination of update and diff. Without this we can only see what was processed, but not what was sent. |
skiplabsdaniel
force-pushed
the
monitoring
branch
8 times, most recently
from
c0909fe
to
e0de197
Compare
skiplabsdaniel
force-pushed
the
monitoring
branch
5 times, most recently
from
273bfd0
to
127c306
Compare
|
On the client side, I think we need to allow it with options for sending data to ours or/and app developer monitoring servers. |
skiplabsdaniel
force-pushed
the
monitoring
branch
4 times, most recently
from
aa40a35
to
823b7c0
Compare
skiplabsdaniel
force-pushed
the
monitoring
branch
from
823b7c0
to
0118b75
Compare
skiplabsdaniel
force-pushed
the
monitoring
branch
2 times, most recently
from
975b322
to
385cef5
Compare
skiplabsdaniel
force-pushed
the
monitoring
branch
from
385cef5
to
49922f1
Compare
skiplabsdaniel
force-pushed
the
monitoring
branch
2 times, most recently
from
8dc9fa0
to
361dcbd
Compare
skiplabsdaniel
force-pushed
the
monitoring
branch
3 times, most recently
from
ca4b08e
to
e382572
Compare
- Manage collector json log file - Manage uptrace dev endpoint - Add Uptrace utils - Manage SKDB collector metric - Manage SKDB uptrace dashboard
skiplabsdaniel
force-pushed
the
monitoring
branch
from
e382572
to
ab50d6b
Compare
No description provided.