Revert to long polling in the ServalClientHelper. Fixes #98.

[Enkidu93]

I think this is the intended behavior. The only change I'm realizing I should possibly make is to make the Exception more specific (?).

---

This change is 

[johnml1135]

tests/Serval.E2ETests/ServalClientHelper.cs line 132 at r1 (raw file):

                revision++;
            }
            catch

The only thing we should be allowing this to continue on is timeouts - can you check the status code? Will the exception type be different? If it is another error, we should just rethrow.

[johnml1135]

Reviewed 1 of 1 files at r1, all commit messages.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @ddaspit and @Enkidu93)

[Enkidu93]

tests/Serval.E2ETests/ServalClientHelper.cs line 132 at r1 (raw file):

                revision++;
            }
            catch

The only thing we should be allowing this to continue on is timeouts - can you check the status code? Will the exception type be different? If it is another error, we should just rethrow.

Yep, my apologies. I can do that.

[johnml1135]

tests/Serval.E2ETests/ServalClientHelper.cs line 132 at r1 (raw file):

Previously, Enkidu93 (Eli C. Lowry) wrote…

Yep, my apologies. I can do that.

if (status_ == 408)
{
  string responseText_ = ( response_.Content == null ) ? string.Empty : await response_.Content.ReadAsStringAsync().ConfigureAwait(false);  throw new ServalApiException("The long polling request timed out.  Did you start the build?", status_, responseText_, headers_, null);
}

[ddaspit]

Reviewed 1 of 1 files at r2, all commit messages.
Reviewable status: all files reviewed, 2 unresolved discussions (waiting on @Enkidu93)

---

tests/Serval.E2ETests/ServalClientHelper.cs line 130 at r2 (raw file):

                    break;
                }
                revision++;

We want to use the revision from the returned build model.

[johnml1135]

Reviewed 1 of 1 files at r2, all commit messages.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @Enkidu93)

[ddaspit]

Reviewed 1 of 1 files at r3, all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @Enkidu93)

[Enkidu93]

tests/Serval.E2ETests/ServalClientHelper.cs line 132 at r1 (raw file):

Previously, johnml1135 (John Lambert) wrote…

if (status_ == 408)
{
  string responseText_ = ( response_.Content == null ) ? string.Empty : await response_.Content.ReadAsStringAsync().ConfigureAwait(false);  throw new ServalApiException("The long polling request timed out.  Did you start the build?", status_, responseText_, headers_, null);
}

Ultimately, this is what is thrown here. Although it's a bit janky, checking based on the message ought to be reliable. Is this OK with you, John?

[johnml1135]

tests/Serval.E2ETests/ServalClientHelper.cs line 132 at r1 (raw file):

Previously, Enkidu93 (Eli C. Lowry) wrote…

Ultimately, this is what is thrown here. Although it's a bit janky, checking based on the message ought to be reliable. Is this OK with you, John?

The ServalApiException has a StatusCode - can you check that instead?

[johnml1135]

Reviewed all commit messages.
Reviewable status: 0 of 1 files reviewed, all discussions resolved (waiting on @ddaspit)

[johnml1135]

tests/Serval.E2ETests/ServalClientHelper.cs line 129 at r4 (raw file):

                }
                revision = result.Revision;
            }

revision = result.Revision + 1; // look for the next revision, the current one

[johnml1135]

Reviewed 1 of 1 files at r4.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @Enkidu93)

[ddaspit]

Reviewed 1 of 1 files at r4, all commit messages.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @johnml1135)

---

tests/Serval.E2ETests/ServalClientHelper.cs line 129 at r4 (raw file):

Previously, johnml1135 (John Lambert) wrote…

revision = result.Revision + 1; // look for the next revision, the current one

He is passing revision + 1 to GetBuildAsync.

[Enkidu93]

tests/Serval.E2ETests/ServalClientHelper.cs line 132 at r1 (raw file):

Previously, johnml1135 (John Lambert) wrote…

The ServalApiException has a StatusCode - can you check that instead?

I tried that initially, but that's not actually what gets thrown here. There are a few timeouts at play here. As I understand it, this timeout is actually the timeout thrown by the HTTP client itself. I am realizing though that this timeout is configured to only 10 seconds. Should we reconfigure it to be greater than the longpolltimeout option?

[Enkidu93]

Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @johnml1135)

---

tests/Serval.E2ETests/ServalClientHelper.cs line 129 at r4 (raw file):

Previously, ddaspit (Damien Daspit) wrote…

He is passing revision + 1 to GetBuildAsync.

If it would be clearer to do otherwise, that's fine. But, of course, it has the same effect.

[ddaspit]

Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @johnml1135)

---

tests/Serval.E2ETests/ServalClientHelper.cs line 129 at r4 (raw file):

Previously, Enkidu93 (Eli C. Lowry) wrote…

If it would be clearer to do otherwise, that's fine. But, of course, it has the same effect.

I don't have an opinion either way.

[johnml1135]

tests/Serval.E2ETests/ServalClientHelper.cs line 129 at r4 (raw file):

Previously, ddaspit (Damien Daspit) wrote…

I don't have an opinion either way.

I didn't see it - it is clear enough.

[johnml1135]

tests/Serval.E2ETests/ServalClientHelper.cs line 132 at r1 (raw file):

Previously, Enkidu93 (Eli C. Lowry) wrote…

I tried that initially, but that's not actually what gets thrown here. There are a few timeouts at play here. As I understand it, this timeout is actually the timeout thrown by the HTTP client itself. I am realizing though that this timeout is configured to only 10 seconds. Should we reconfigure it to be greater than the longpolltimeout option?

Absolutely - we would want it to present itself as a proper message. We could also change it by having the long-poll timeout be under the http timeout, say 9 seconds? @ddaspit what do you think?

[johnml1135]

Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @Enkidu93)

[ddaspit]

Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @Enkidu93)

---

tests/Serval.E2ETests/ServalClientHelper.cs line 132 at r1 (raw file):

Previously, johnml1135 (John Lambert) wrote…

Absolutely - we would want it to present itself as a proper message. We could also change it by having the long-poll timeout be under the http timeout, say 9 seconds? @ddaspit what do you think?

Yes, we should set the request timeout for HttpClient to something higher than the long polling timeout. If we want a shorter timeout for other requests, we can control that using the cancellation token.

[Enkidu93]

tests/Serval.E2ETests/ServalClientHelper.cs line 132 at r1 (raw file):

Previously, ddaspit (Damien Daspit) wrote…

Yes, we should set the request timeout for HttpClient to something higher than the long polling timeout. If we want a shorter timeout for other requests, we can control that using the cancellation token.

I went ahead and updated the timeout and the catch here. Question though: Is the Task.Delay even necessary at this point?

[johnml1135]

Task delay is still necessary if it increments very quickly. I have seen dozens if not, hundreds of requests per second on SMT training. There should be a max rate for updating. Get Outlook for iOS<https://aka.ms/o0ukef>

…

________________________________ From: Eli C. Lowry ***@***.***> Sent: Friday, September 29, 2023 4:14:14 PM To: sillsdev/serval ***@***.***> Cc: John Lambert ***@***.***>; Mention ***@***.***> Subject: Re: [sillsdev/serval] Revert to long polling in the ServalClientHelper. Fixes #98. (PR #143) tests/Serval.E2ETests/ServalClientHelper.cs line 132 at r1<https://reviewable.io/reviews/sillsdev/serval/143#-NfM2tV97d8QAMr7n14w:-NfXZn6q-aiQ4ZP8BcGh:b-8e2gn> (raw file<https://github.com/sillsdev/serval/blob/e0829dd3952465565df428a6455d4aadd088fbda/tests/Serval.E2ETests/ServalClientHelper.cs#L132>): Previously, ddaspit (Damien Daspit) wrote… Yes, we should set the request timeout for HttpClient to something higher than the long polling timeout. If we want a shorter timeout for other requests, we can control that using the cancellation token. I went ahead and updated the timeout and the catch here. Question though: Is the Task.Delay even necessary at this point? — Reply to this email directly, view it on GitHub<#143 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ADIY5NFKDXSE3H5VTBHG5DLX44TZNANCNFSM6AAAAAA5IJHRUI>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[Enkidu93]

Task delay is still necessary if it increments very quickly. I have seen dozens if not, hundreds of requests per second on SMT training. There should be a max rate for updating. Get Outlook for iOShttps://aka.ms/o0ukef
…
________________________________ From: Eli C. Lowry @.> Sent: Friday, September 29, 2023 4:14:14 PM To: sillsdev/serval @.> Cc: John Lambert @.>; Mention @.> Subject: Re: [sillsdev/serval] Revert to long polling in the ServalClientHelper. Fixes #98. (PR #143) tests/Serval.E2ETests/ServalClientHelper.cs line 132 at r1https://reviewable.io/reviews/sillsdev/serval/143#-NfM2tV97d8QAMr7n14w:-NfXZn6q-aiQ4ZP8BcGh:b-8e2gn (raw file

serval/tests/Serval.E2ETests/ServalClientHelper.cs

Line 132 in e0829dd

catch

): Previously, ddaspit (Damien Daspit) wrote… Yes, we should set the request timeout for HttpClient to something higher than the long polling timeout. If we want a shorter timeout for other requests, we can control that using the cancellation token. I went ahead and updated the timeout and the catch here. Question though: Is the Task.Delay even necessary at this point? — Reply to this email directly, view it on GitHub<#143 (comment)>, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ADIY5NFKDXSE3H5VTBHG5DLX44TZNANCNFSM6AAAAAA5IJHRUI. You are receiving this because you were mentioned.Message ID: @.***>

OK, great! Can you approve this request? Or is there something left to do?

[johnml1135]

Reviewed 1 of 1 files at r5, all commit messages.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @ddaspit)

[johnml1135]

Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @ddaspit)

[ddaspit]

I agree that we update the status too much for SMT training. Do we have an issue for that? We should throttle those updates internally.

Reviewable status: complete! all files reviewed, all discussions resolved