Loading reactor very long time

Moderator: SecondMan

User avatar
malbred
Posts: 24
Joined: Wed Aug 16, 2017 9:17 pm
Location: Saint-Petersburg, Russia
Contact:

Loading reactor very long time

#1

Post by malbred » Thu Apr 26, 2018 7:01 am

Recently I have a very long time reactor loading. More than 20 minutes.
During this time, I see the initialization, update icons, etc. Without any errors
What could it be? What i must to check?
F 9.02
Image

User avatar
AndrewHazelden
Fusionator
Posts: 1001
Joined: Fri Apr 03, 2015 3:20 pm
Location: West Dover, Nova Scotia, Canada
Been thanked: 91 times
Contact:

Re: Loading reactor very long time

#2

Post by AndrewHazelden » Thu Apr 26, 2018 7:35 am

malbred wrote:
Thu Apr 26, 2018 7:01 am
Recently I have a very long time reactor loading. More than 20 minutes.
During this time, I see the initialization, update icons, etc. Without any errors
What could it be? What i must to check?
Hi @malbred.

The slower then normal speed is caused by the GitLab site that hosts the Reactor repository for the atoms. I've noticed this issue over the last two days.

You can see a notice about the GitLab issues on their Twitter feed here:




More details on the GitLab database issues that are causing the slowdowns are listed here if you are curious:

https://docs.google.com/document/d/1_Iz ... Zw/preview

User avatar
SirEdric
Fusionista
Posts: 902
Joined: Tue Aug 05, 2014 10:04 am
Been thanked: 62 times
Contact:

Re: Loading reactor very long time

#3

Post by SirEdric » Thu Apr 26, 2018 9:50 am

Couldn't @SecondMan just set up our very own WSLgit on this very server....:-)

User avatar
AndrewHazelden
Fusionator
Posts: 1001
Joined: Fri Apr 03, 2015 3:20 pm
Location: West Dover, Nova Scotia, Canada
Been thanked: 91 times
Contact:

Re: Loading reactor very long time

#4

Post by AndrewHazelden » Thu Apr 26, 2018 10:55 am

SirEdric wrote:
Thu Apr 26, 2018 9:50 am
Couldn't @SecondMan just set up our very own WSLgit on this very server....:-)
If the Reactor git repository was self hosted it would dramatically increase hosting costs and IT overhead so that wouldn't be prudent for a 100% volunteer run project. :)

Reactor v1.1 is adding support for both Fusion Standalone and Resolve 15 installation so the bandwidth usage levels will likely spike a lot higher when the potential userbase for the Reactor package manager takes an exponential jump higher in size.

Reactor for Fusion will be at 1400 installed users by the end of the month. If Resolve 15 drove that up by 10x or higher it would be lot more load for the web hosting plan to absorb on a free open-source project.

User avatar
SirEdric
Fusionista
Posts: 902
Joined: Tue Aug 05, 2014 10:04 am
Been thanked: 62 times
Contact:

Re: Loading reactor very long time

#5

Post by SirEdric » Thu Apr 26, 2018 12:30 pm

Mate...I was joking...:-)
(But still...the idea is intriguing, innit?....:-))

User avatar
SecondMan
Site Admin
Posts: 2565
Joined: Thu Jul 31, 2014 5:31 pm
Location: Vancouver, Canada
Been thanked: 86 times
Contact:

Re: Loading reactor very long time

#6

Post by SecondMan » Thu Apr 26, 2018 1:06 pm

SirEdric wrote:
Thu Apr 26, 2018 9:50 am
Couldn't @SecondMan just set up our very own WSLgit on this very server....:-)

Hell no :D

I do need some sleep, my dear Eric ;)

And yes, what Andrew said. Also a public, production proven service like GitLab is way more likely to get adopted than some private effort run by goodness-knows-who.

Plus people can fork there to their own accounts, build on it, etc etc.

User avatar
AndrewHazelden
Fusionator
Posts: 1001
Joined: Fri Apr 03, 2015 3:20 pm
Location: West Dover, Nova Scotia, Canada
Been thanked: 91 times
Contact:

Re: Loading reactor very long time

#7

Post by AndrewHazelden » Thu Apr 26, 2018 1:25 pm

SirEdric wrote:
Thu Apr 26, 2018 12:30 pm
Mate...I was joking...:-)
(But still...the idea is intriguing, innit?....:-))
Well it would be *technically* possible to do something like that. :D

Reactor has a "protocols" interface "Reactor:/System/Protocols/*.lua" that allows you to source your atom content from multiple locations stored locally or via a network using several different technologies. Also, multiple concurrent repositories can be configured in the same "Reactor:/System/Reactor.cfg" file.

Extra communication protocols for syncing could be added down the road to Reactor as needed using this "protocols" approach via short Lua scripts.

Fusion has the cURL network library built-in via Lua FFI which means strange and alternative Reactor atom syncing systems like HTTP/HTTPS, WebDAV, FTP/FTPS, SSH, SMB, NFS, Gopher, Torrents, GitHub repos, Dropbox, or Google Drive could theoretically be coded in the distant future if a developer took on the task to code the individual protocols. 8-)

User avatar
SirEdric
Fusionista
Posts: 902
Joined: Tue Aug 05, 2014 10:04 am
Been thanked: 62 times
Contact:

Re: Loading reactor very long time

#8

Post by SirEdric » Thu Apr 26, 2018 8:36 pm

AndrewHazelden wrote:
Thu Apr 26, 2018 1:25 pm
if a developer took on the task to code the individual protocols. 8-)
For some reason I very much like the wording in that last part, combined with the Cool Glasses Smiley...:-)

User avatar
TogTobias
Fusioneer
Posts: 214
Joined: Fri Feb 20, 2015 5:58 pm
Contact:

Re: Loading reactor very long time

#9

Post by TogTobias » Thu May 03, 2018 6:58 am

So. Is there going to be any movement on this? Only asking because this is seriously cramping productivity having to wait several minutes per load.

thibaud
Fusioneer
Posts: 108
Joined: Thu Sep 04, 2014 1:23 am
Been thanked: 1 time

Re: Loading reactor very long time

#10

Post by thibaud » Fri May 04, 2018 12:03 am

I'm not too sure te solution lies in the server infrastructure.
perhaps remote access could be minimized by implementing a caching system

emill
Posts: 4
Joined: Thu Nov 02, 2017 7:31 am
Contact:

Re: Loading reactor very long time

#11

Post by emill » Fri May 25, 2018 1:13 am

Same issue here.
thibaud wrote:
Fri May 04, 2018 12:03 am
I'm not too sure te solution lies in the server infrastructure.
perhaps remote access could be minimized by implementing a caching system
+1. UI should load right away with locally cached data and Reactor should ask the server for updated data in the background.

thibaud
Fusioneer
Posts: 108
Joined: Thu Sep 04, 2014 1:23 am
Been thanked: 1 time

Re: Loading reactor very long time

#12

Post by thibaud » Fri May 25, 2018 2:21 am

I hate to say it, but I can't use reactor at all.
launching time is about 15min... you close it... try to relaunch it... another 15min

User avatar
SecondMan
Site Admin
Posts: 2565
Joined: Thu Jul 31, 2014 5:31 pm
Location: Vancouver, Canada
Been thanked: 86 times
Contact:

Re: Loading reactor very long time

#13

Post by SecondMan » Fri May 25, 2018 7:14 am

Should, should... Welcome to the forum ;)

Yes we are noticing the same. Rest assured we're looking into this, everyone. Unfortunately there's a bit of bad luck involved here. One of the new features of Reactor 2.0 is that it does have a working caching system that updates content only as needed. There are a few small core files left to receive the same treatment plus there's obviously a check on startup for new content with an active internet connection. Those files are so small and our many tests ran so smoothly that we did not consider that step a critical priority.

But, I kid you not - on the very day of the release, GitLab started having major issues with API access and they have been lasting for days now. It's been a while since I've seen 1kb/s download speeds... So we're looking for alternatives there, too, with the obvious choice being GitHub.

It's a shame really, because when GitLab is optimal - the issues are intermittent - Reactor really flies.

So please 🐻 with us and let's for the moment hope that the GitLab issues are only very temporary.

emill
Posts: 4
Joined: Thu Nov 02, 2017 7:31 am
Contact:

Re: Loading reactor very long time

#14

Post by emill » Thu May 31, 2018 11:43 pm

@SecondMan Thank you for the welcome :)

Did not mean to sound like the stereotypical rude/angry/impatient user, could have taken a few more seconds to word my message a little better. I realize you guys are putting serious effort into this and I appreciate it.

Sounds like you realize how important the startup update delay is. To me it seems that ideally any internet-dependent functions of Reactor would be run async and the results would only be utilized in callback in case the server responds. Otherwise it would be all local (except ofc there needs to be a bigger successful download right after install). And maybe there could be a warning icon if it's been too long since the last successful data update.

Dev work isn't easy so thanks again!