passive-yellow•3y ago

Mark session as bad when request times out or proxy responds with 502

I'm using CheerioCrawler and I'd like to mark sessions as bad when the request either times out or there's a proxy error. Those cases trigger an error before reaching requestHandler and the request is added back to the queue without me having the opportunity to mark the session. Is there a hook somewhere that I can use? Or should I override _requestFunctionErrorHandler?

16 Replies

fair-rose•3y ago

I would like to know this as well

fascinating-indigo•3y ago

You can mark a session as bad with the session.markBad() function within the errorHandler function (which runs on every request failed, as opposed to failedRequestHandler, which runs once a request has reached its max retries)

const crawler = new CheerioCrawler({
    proxyConfiguration,
    requestHandler: router,
    errorHandler: ({ session }) => {
        session.markBad();
    },
});

const crawler = new CheerioCrawler({
    proxyConfiguration,
    requestHandler: router,
    errorHandler: ({ session }) => {
        session.markBad();
    },
});

But if you just want a session to be thrown away if it fails once, you can do this instead in the sessionPoolOptions:

const crawler = new CheerioCrawler({
    proxyConfiguration,
    requestHandler: router,
    sessionPoolOptions: {
        sessionOptions: {
            maxErrorScore: 1,
        },
    },
});

const crawler = new CheerioCrawler({
    proxyConfiguration,
    requestHandler: router,
    sessionPoolOptions: {
        sessionOptions: {
            maxErrorScore: 1,
        },
    },
});

passive-yellowOP•3y ago

Amazing thank you @thek1tten I didn't know about errorHandler One more question: how can I access the error in errorHandler? Is it passed as parameter? All good I found my answer in the docs! @thek1tten can I prevent the request from being retried depending on the error from the errorHandler?

fascinating-indigo•3y ago

fascinating-indigo•3y ago

This should work

passive-yellowOP•3y ago

So I've tried that but without success, the request still ends up being retried Is there any other way to prevent a retry? Maybe throwing a NonRetryableError?

passive-yellowOP•3y ago

See on the logs, I print request right after setting request.noRetry to true in errorHandler, then the request is retried right after

fascinating-indigo•3y ago

Hmm, that means it’s going off of the old value and reassigning it here does nothing. Let me look into it.

passive-yellowOP•3y ago

Thanks!

MEE6•3y ago

@fab8203 just advanced to level 4! Thanks for your contributions! 🎉

fascinating-indigo•3y ago

This feature doesn’t seem to exist yet. I’m making a PR on Crawlee’s GitHub to fix this

passive-yellowOP•3y ago

Thank you @thek1tten let me know if there is a link to the issue that I can follow

fascinating-indigo•3y ago

https://github.com/apify/crawlee/pull/1542

GitHub

feat(basic-crawler): allow request skipping by mstephen19 · Pull Re...

See this Discord post to fully understand the use case: https://discord.com/channels/801163717915574323/1019936393235017769 Didn't want to make big changes to existing code so kept the else sta...

fascinating-indigo•3y ago

@fab8203 It was merged with master

passive-yellowOP•3y ago

Thank you for the follow up @thek1tten

Lukas Krivka•3y ago

Btw: retiring sessions and not retrying request are 2 completely different concepts. Request and Session are separate objects that might be connected temporarily (do a Request using this Session)

Mark session as bad when request times out or proxy responds with 502

Did you find this page helpful?