57 comments

[ 4.0 ms ] story [ 122 ms ] thread
Starting about 30 minutes ago, my Slack clients all started having issues connecting to Slack. It went from working to not working in about 10 seconds. I've also been trying to access the Slack Status page with little success. Their Twitter account doesn't seem to be very helpful, but is at least working: https://twitter.com/SlackHQ
Thus begins the slew of "Slack oh noes" posts.
Actually there have been more posts like yours, predicting those types of posts than those types of posts themselves.
Interesting, because I saw only mine, but 6-12 Slack posts.
We (https://sameroom.io) suggest rigging up a contingency chat solution for situations like these.

Our service can be used to keep key channels synchronized between main and secondary chat systems—to avoid losing context when switching over. Having this in place is particularly important if _your_ outage coincides with team chat outage.

Here's a post we wrote on this topic: https://sameroom.io/blog/when-team-chat-goes-down-are-you-re...

We'll know all the competitor names in this thread soon
Speaking of, check out https://fleep.io — great product, slightly different model than Slack.
It would be great if it supported linux or at least some client like pidgin
Ditto when AWS or Github go down, as well as everyone who points out why it was obvious to not Slack|Github|AWS. Then the service comes back up and business as usual.
I'm 100% OK with Slack, even through all this, but it seriously needs to be decentralized a bit. I'd like to see pricing on an on-prem version of it.
If you were going to run an on-prem version of Slack why not go with one of the existing tools that does the same thing?
I think because people like using Slack and, for an office communication tool, that's pretty important.
I don't understand why people can't just open up the window and yell to each other, like my impression of New York City in the early 1900s, but hipper and slicker.

Call it windowcasting maybe.

A lighthearted joke to hold off the panic while slack gets their servers back up.

Ever hear a New York Echo? No! Heeeeeellloooooooooooooooooo!!!!! ... ... STFU!!

(comment deleted)
I can't remember the last widespread outage for a major IRC network.

Just sayin'.

On freenode or alike, netsplit is too common, although few would consider that as an outage of any sort.
Reactiflux - the React.js chat/discussion room moved away from Slack to Discord (https://discordapp.com/) a couple of months back. the reason was: "Slack decided that Reactiflux had too many members and disabled new invites"

here is a post with more info: https://facebook.github.io/react/blog/2015/10/19/reactiflux-...

How is this related to the outage?
It is related in the way that there are other tools out there that users can use, if people are frustrated by these outages.
All I can say is I sympathise and empathise with their operations team.

Outages happen to everyone, including I'm sure Discord, and what matters most to me is transparency during and after the incident and that both individual and corporate values include incremental improvement - not making the same mistake twice.

I'm perpetually frustrated by HN comments which take a tack of either pointing out (with your benefit of hindsight) things the commenter feels were incorrect, professing they could do better, or the strong advocacy for abandoning ship even with companies who have a provably pretty decent uptime record.

I am also frustrated with demoting comments that offer other alternatives or paint a different side of the picture.

Unfortunately I cannot sympathize with a company which it's valuation is $2.8 billon, and cannot handle staying up or a large userbase.

I didn't downvote you, but I suspect the reason you were is because it smells like opportunistic astroturf.
Better don't mess with Reactiflux?
Thankfully their status page[1] is back up, though it was flaking for a bit earlier. Transparency is the most important thing to me.

The status page is what I use to update StatusGator[2] so it can notify you when services go down. It can even notify you by Slack (when it's up) and for that reason I recommend using a second mode as well such as email, SMS, or webhook if you want to get crafty.

1. https://status.slack.com

2. https://statusgator.io

Meanwhile, IRC continues to work exactly as well as it has for 20+ years.
IRC servers go down, too
You're the third person I've seen make this remark. Are you trying to be annoying or are you really missing the point this extremely?
Thats why you have multiple hubs and multiple client servers....redundancy!
IRC is decentralized, so a few severs going down is no problem.
netsplit isn't exactly what I'd call "no problem".
I suspect most internal IRC nets just consist of a single server
That's not exactly a point for slack, since internal slack nets don't exist. I call this "downtime as a service": you get all of the satisfaction an interrupted work day without any of the annoyance of being able to fix it yourself.
The flip side of that is with Slack, you have all the same problems as when your internal IRC server going down, but without the additional productivity loss of internal resources having to drop what they're doing to fix the damn thing.
I wasn't being 100% sarcastic. It's sometimes actually nice to use saas downtime as an excuse to take the day off without being the guy who has to fix it.

    sudo service ircd-irc2 restart
I'm pretty sure SSH-ing into the server and executing that command costs less in productivity than waiting for Slack to fix their issues.
A netsplit is just as annoying
Yeah, I don't see a need to have more than one server. Even at gigantic companies, you only need single servers per team. It takes very little to restart a server:

    sudo service ircd-irc2 restart
Meh, I'd rather have one server, honestly. It takes two seconds to restart an IRC server.
If you run your own, hardly. If theres a netsplit, it just means you'll have to wait a second until you're connected again.

But really, if you run your own IRC server in your own internal network, there will be no netsplits.

Instead, it goes down just like Slack.
But you can just go ahead and restart them, and they don't cost $5+ per person per month.

    sudo service ircd-irc2 restart
Fixed it.
Indeed. It is frustrating that my co-workers have gotten so pissy when I asked them to fall back to the internal IRC server because Slack was so flakey.
"Exactly as well" meaning going down fairly often?
Not sure if this is sarcasm or not ;)
Not sarcasm. I'm entirely annoyed with the software industry's constant attempts to make money off problems that were solved by someone else decades ago.

I'm pretty sure this recent instability won't make Slack go out of style. What will likely happen is Slack will continue releasing "features" that don't improve upon IRC and make their product harder to use, until someone else comes along with a simple IRC clone and unseats them just like Slack unseated HipChat. It's an exercise in pointlessness.

Someone asked me just a couple days ago, why don't use use Slack instead of IRC?

This. IRC just works. It's not a walled garden.

Well IRC is able to go down just like Slack, especially if you're using a smaller/private network (I'm not going to run my organisation's 'Slack' on Freenode).

Again, as these things are, the tech has nothing to do with it. Hipchat 'lost' to Slack while being more or less identical (from a raw features/technical point of view). Hell, Slack could be IRC for all I know, but what makes people use Slack is more than what protocol it uses underneath.

> Hell, Slack could be IRC for all I know, but what makes people use Slack is more than what protocol it uses underneath.

Yeah, it's a combination of slavishly always having to be on the next new thing, and being unable to recognize that this is a problem.

I might be alone on this but I find it strange that "XYZ Website is DOWN!!!" is considered news worthy. I mean, if it's down and I care I know it because... it's down. If I don't use it, why do I care again?

But then again, I never read tech crunch anymore because of different but identical crap.

I have to agree. When it's something big and business-critical, like AWS or Azure or some other major host, that's newsworthy, but honestly we should expect something like Slack to go down occasionally. So we can't chat for a few hours on a Monday, email and other business-critical lines of communication are still working just fine.

On the other hand, I don't particularly mind it, either. It's not hurting anything, and maybe we need an excuse to dig up the old "proprietary communication vs IRC" argument now and then, just for kicks.

Slack has become business critical at many companies - quite possibly including TechCrunch (lots of media companies with distributed journalists use Slack to co-ordinate everything).
On HN we penalize such articles, plus users flag them heavily.
Meanwhile, Hipchat has been having issues on and off for weeks and nobody says a thing.