This may be more a general PG question, but asking here in case there's any HD implications/angle on

This may be more a general PG question, but asking here in case there's any HD implications/angle on it: is it advisable to cache the

Client

Client

instance on

req

req

, rather than create it for each query? i.e. instead of:

const runQ = async q => {
    const client = new Client(/* options */);
    await client.connect();
    return client.query(q);
}

const runQ = async q => {
    const client = new Client(/* options */);
    await client.connect();
    return client.query(q);
}

const runQ = async q => {
    const client = new Client(/* options */);
    await client.connect();
    return client.query(q);
}

const runQ = async q => {
    const client = new Client(/* options */);
    await client.connect();
    return client.query(q);
}

do:

const runQ = async (req, q) => {
    if (!req.client) {
        req.client = new Client(/* options */);
        await req.client.connect();
    }
    return req.client.query(q);
}

const runQ = async (req, q) => {
    if (!req.client) {
        req.client = new Client(/* options */);
        await req.client.connect();
    }
    return req.client.query(q);
}

const runQ = async (req, q) => {
    if (!req.client) {
        req.client = new Client(/* options */);
        await req.client.connect();
    }
    return req.client.query(q);
}

const runQ = async (req, q) => {
    if (!req.client) {
        req.client = new Client(/* options */);
        await req.client.connect();
    }
    return req.client.query(q);
}

MMitya This may be more a general PG question, but asking here in case there's any HD i...

Isaac McFadyen•8/16/24, 12:49 PM

No.

req

req

is new every time and the client will be recycled.

Isaac McFadyen•8/16/24, 12:49 PM

Also it's fine to recreate clients with Hyperdrive, it holds open connections for you.

IIsaac McFadyen Also it's fine to recreate clients with Hyperdrive, it holds open connections fo...

Hard@Work•8/16/24, 1:18 PM

I think they meant if a single Request runs multiple queries across different handlers

Isaac McFadyen•8/16/24, 1:18 PM

Oh I see

Isaac McFadyen•8/16/24, 1:18 PM

Then yeah, could work. I'd still be wary of mutating request state though... I've always been told not to do that by various frameworks

Isaac McFadyen•8/16/24, 1:19 PM

If Workers changes something internally on the request you have no guarantee that

req.client

req.client

will still be unused (unlikely it will be but possible?)

IIsaac McFadyen Then yeah, could work. I'd still be wary of mutating request state though... I'v...

Hard@Work•8/16/24, 1:20 PM

Some frameworks do provide a "state" store somewhere that will be passed around, so I would use that if possible

HHard@Work Some frameworks do provide a "state" store somewhere that will be passed around,...

Isaac McFadyen•8/16/24, 1:20 PM

Yeah this

Isaac McFadyen•8/16/24, 1:20 PM

Hono calls it context, Fastify calls it state.

Isaac McFadyen•8/16/24, 1:20 PM

There's nothing workers-native though I don't think.

IIsaac McFadyen There's nothing workers-native though I don't think.

Hard@Work•8/16/24, 1:35 PM

Isn't the "official" one

req.cf

req.cf

Hard@Work•8/16/24, 1:36 PM

Like, as in, they don't guarantee that they won't put new stuff on it, but it will be passed around

HHard@Work Isn't the "official" one `req.cf`?

Isaac McFadyen•8/16/24, 1:36 PM

Yeah but that's even more likely to be overwritten.

Isaac McFadyen•8/16/24, 1:36 PM

I guess if you name your thing something obscure it'll work, but I can imagine a

req.cf.client

req.cf.client

some day for some service or another.

moody 💭•8/16/24, 4:46 PM

seem to be experiencing an outage again, with hyperdrive connection latency spikes. is there anyone available who can look into it?

AJR•8/16/24, 4:49 PM

Let me take a look. Same config ID you'd previously sent?

AJR•8/16/24, 5:09 PM

Looks like the ID you'd sent previously is not seeing traffic. Please post or DM the one you're seeing issues with. I'm not seeing latency spikes globally, but individual configs certainly could be. Happy to take a look.

AAJR Looks like the ID you'd sent previously is not seeing traffic. Please post or D...

moody 💭•8/16/24, 7:11 PM

dmd

IIsaac McFadyen Oh I see

MityaOP•8/19/24, 11:57 AM

Interesting - I've seen frameworks absolutely encourage storing things on

req

req

between middleware. Otherwise you'd have no option but to laboriously pass around manually a ton of data as function arguments.

HHard@Work I think they meant if a single Request runs multiple queries across different ha...

MityaOP•8/19/24, 12:01 PM

Exactly.

MMitya Interesting - I've seen frameworks absolutely encourage storing things on `req` ...

Isaac McFadyen•8/19/24, 12:15 PM

Absolutely, but there's usually a property dedicated towards it.

Isaac McFadyen•8/19/24, 12:15 PM

ctx

ctx

or similar.

Ping for toast•8/19/24, 9:47 PM

https://www.cloudflarestatus.com/incidents/035gqnkz86zh
was there possibly downtime related to this

Elevated 1xxx Error Rates

PPing for toast https://www.cloudflarestatus.com/incidents/035gqnkz86zh was there possibly downt...

AJR•8/19/24, 9:49 PM

No, not for Hyperdrive. Not that I've seen or know of.

AAJR No, not for Hyperdrive. Not that I've seen or know of.

Ping for toast•8/19/24, 9:50 PM

odd. I know that they have the same ip pool so i was wondering because for about a minute i had connection timeouts

bs_sfalls•8/19/24, 11:21 PM

Seeing an interesting problem with workers, may or may not be a result of hyperdrive. I have an API implemented in workers with typescript and I'm running React in Pages. Some of my components load 2 different requests simultaneously. Sometimes, works fine, a lot of times, one of the requests receives a connection timeout from the DB. If I disable hyperdrive, the problem goes away. If I stagger the request by a second or two then there is no problem. Is it possible hyperdrive or workers cannot locate a connection in such quick succession? Should I serialize the requests?

Interestingly, I also see the problem locally with wrangler dev. Again, stagger the requests, no issue.

Bbs_sfalls Seeing an interesting problem with workers, may or may not be a result of hyperd...

Matt Silverlock•8/20/24, 12:29 AM

If you’re seeing it in dev - where Hyperdrive’s connection logic isn’t in place - it might be something else. Are you doing something odd with your driver/client code?

MMatt Silverlock If you’re seeing it in dev - where Hyperdrive’s connection logic isn’t in place ...

bs_sfalls•8/20/24, 5:22 PM

Nothing unique and no problems normally. The only issue is if I send off 2 simultaneous connections from react at roughly the same time. I see it in wrangler and I see it in workers production. Using Hono and Kysley. I'll dig a bit more, as of right now, just staggering the requests.

Bbs_sfalls Nothing unique and no problems normally. The only issue is if I send off 2 simu...

Ping for toast•8/20/24, 9:07 PM

I think its possible I'm experiencing this as well

PPing for toast odd. I know that they have the same ip pool so i was wondering because for about...

Ping for toast•8/20/24, 9:08 PM

up here

Ping for toast•8/20/24, 9:08 PM

started getting this error a couple days ago

Ping for toast•8/20/24, 9:09 PM

that looks like a bug

Ping for toast•8/20/24, 9:09 PM

and i suppose it is

Ping for toast•8/20/24, 9:09 PM

but whats interesting is

Ping for toast•8/20/24, 9:11 PM

function n(i) {
  e50.apply(null, i), r = null;
}
function un() {
  Et($r.connection("CONNECT_TIMEOUT", e50, D)), D.destroy();
}
function Et(H) {
  ae && (ae.destroy(H), ae = null), ye && ft(ye, H), z && (ft(z, H), z = null);
}
function ft(H, ue) {
  Object.defineProperties(ue, { stack: { value: ue.stack + H.origin.replace(/.*\n/, `
`), enumerable: e50.debug }, query: { value: H.string, enumerable: e50.debug }, parameters: { value: H.parameters, enumerable: e50.debug }, args: { value: H.args, enumerable: e50.debug }, types: { value: H.statement && H.statement.types, enumerable: e50.debug } }), H.reject(ue);
}

function n(i) {
  e50.apply(null, i), r = null;
}
function un() {
  Et($r.connection("CONNECT_TIMEOUT", e50, D)), D.destroy();
}
function Et(H) {
  ae && (ae.destroy(H), ae = null), ye && ft(ye, H), z && (ft(z, H), z = null);
}
function ft(H, ue) {
  Object.defineProperties(ue, { stack: { value: ue.stack + H.origin.replace(/.*\n/, `
`), enumerable: e50.debug }, query: { value: H.string, enumerable: e50.debug }, parameters: { value: H.parameters, enumerable: e50.debug }, args: { value: H.args, enumerable: e50.debug }, types: { value: H.statement && H.statement.types, enumerable: e50.debug } }), H.reject(ue);
}

Ping for toast•8/20/24, 9:11 PM

CONNECT_TIMEOUT?

Ping for toast•8/20/24, 9:16 PM

function connectTimedOut() {
  errored(Errors.connection('CONNECT_TIMEOUT', options, socket))
  socket.destroy()
}
function errored(err) {
  stream && (stream.destroy(err), stream = null)
  query && queryError(query, err)
  initial && (queryError(initial, err), initial = null)
}
function queryError(query, err) {
  Object.defineProperties(err, {
    stack: { value: err.stack + query.origin.replace(/.*\n/, '\n'), enumerable: options.debug },
    query: { value: query.string, enumerable: options.debug },
    parameters: { value: query.parameters, enumerable: options.debug },
    args: { value: query.args, enumerable: options.debug },
    types: { value: query.statement && query.statement.types, enumerable: options.debug }
  })
  query.reject(err)
}

function connectTimedOut() {
  errored(Errors.connection('CONNECT_TIMEOUT', options, socket))
  socket.destroy()
}
function errored(err) {
  stream && (stream.destroy(err), stream = null)
  query && queryError(query, err)
  initial && (queryError(initial, err), initial = null)
}
function queryError(query, err) {
  Object.defineProperties(err, {
    stack: { value: err.stack + query.origin.replace(/.*\n/, '\n'), enumerable: options.debug },
    query: { value: query.string, enumerable: options.debug },
    parameters: { value: query.parameters, enumerable: options.debug },
    args: { value: query.args, enumerable: options.debug },
    types: { value: query.statement && query.statement.types, enumerable: options.debug }
  })
  query.reject(err)
}

its a bug in postgres.js, caused by CONNECTION_TIMEOUTs

Bbs_sfalls Nothing unique and no problems normally. The only issue is if I send off 2 simu...

Ping for toast•8/20/24, 9:38 PM

Furthermore, I can corroborate that the issue is solved when hyperdrive is disabled

Ping for toast•8/20/24, 9:39 PM

I can tell that hyperdrive was properly disabled because requests now take about 200ms longer

AJR•8/20/24, 10:04 PM

How interesting. Can you share a code snippet that induces this behavior? How often/reliably does it reproduce for you?

AJR•8/20/24, 10:06 PM

Aside, you should be seeing better caching behavior now, btw, @Ping for toast . How's it looking on your side?

AAJR How interesting. Can you share a code snippet that induces this behavior? How ...

Ping for toast•8/20/24, 10:11 PM

By a code snippet, do you mean you're asking for a minimum reproducible example? My configuration is very much like bs_sfalls

I use solid-start and hyperdrive, and my backend is elysia with hyperdrive.
The backend is hosted on a separate worker.
In SSR, two requests are fired simultaneously — one to retrieve data about the current user, and one to fetch the data presented on the page, using service bindings
This can obviously be triggered by doing a page reload.
Until said page reload happens, requests work fine
After n seconds (where n is unknown; greater than 30s and less than 5 minutes?), presumably after hyperdrive disconnects // the worker cools down, it begins working again

It seems to reproduce perfectly.

Ping for toast•8/20/24, 10:11 PM

Caching is a lot better — 50% or so

Ping for toast•8/20/24, 10:12 PM

I also literally just switched db hosts as I thought this would be a good opportunity, so I will test it again to see if it reproduces

AJR•8/20/24, 10:16 PM

Interesting. We do have some rate limiting in place but it wouldn't be triggered by a single pair of simultaneous requests, and even if it was it would only lock out for a couple hundred ms.

I'm going to have to dig at this a bit. I'll put together a ticket for it

You said this only started within the last week or so?

@bs_sfalls same timeframe for you, just the last week or so?

AAJR Interesting. We do have some rate limiting in place but it wouldn't be trigger...

Ping for toast•8/20/24, 10:16 PM

yup, you can actually see the first time i noticed it

PPing for toast https://www.cloudflarestatus.com/incidents/035gqnkz86zh was there possibly downt...

Ping for toast•8/20/24, 10:16 PM

right here :)

Ping for toast•8/20/24, 10:16 PM

it could have been up to a week tho

Ping for toast•8/20/24, 10:23 PM

Reproduced

Ping for toast•8/20/24, 10:24 PM

The reason that it looks like the html has loaded is because of streaming

This may be more a general PG question, but asking here in case there's any HD implications/angle on

Similar Threads