Network lag/issues Megathread

Demokdawa, this is not necessarily related to low bandwidth. I have a 1.2Gbp/s connection (ok, on wifi it’s closer to like 100Mb/s). Tracking all of this could really help rule out the extraneous factors from the common ones. Some people mention necro + minions, I’ve had this happen on spellblade with no minions as well as necro, etc.

But anyway, devs have to prioritize, and maybe this isn’t worth the effort given the current impact. We don’t have that kind of visibility unfortunately.

WE don’t have that kind of visibility, but i Nevers Seen someone says that they play at 3 or 4 players without lags.

I have personnaly no issues solo with my bandwith (same as you), and it’s the same for every Friend with good bandwith i have thats play LE.

Only Friends that get issues in solo are thé one with low download/upload speeds.

But when you play as a group, when thé lag become too high, everyone get desyncs at thé same Time, for thé same duration, etc… 3p and 4p doesn’t Care about bandwith, it just lags for everyone the same.

When in play minions projectile builds, only thé Friends (in 2p) with low bandwith lags out. I Nevers get affected by their builds whatsoever.

Are we sure it’s the original Forge they’re using, and not Forge Alloy? If it is indeed the original Forge, at least that should be an easy change for them to hop over to Forge Alloy, which theoretically addresses at least one of the concerns (If I read the commits properly)

Somewhere in ILSPY:

// BeardedManStudios.Forge.Networking.BasePacketComposer
using BeardedManStudios.Forge.Networking.Frame;
...

In ForgeAlloy unit test reference this way:

using Forge.Networking.Messaging;

In BeardedManStudios ForgeNetworking

using BeardedManStudios.Forge.Networking;

We can assume it’s the BeardedManStudios version.

Some tests I’ve done:

  1. Hopping around different VPN servers seems to have slightly different behavior and sometimes alleviates how often the desync happens. Seems very inconsistent for me though.
  2. Only my pc running Last Epoch has general network degradation whenever Last Epoch does its thing. Laptop on the same wifi sees perfectly normal performance.
  3. Killing Last Epoch immediately sees the rest of the network recover.

Before 3) I suspected maybe my ISP was doing something after detecting p2p traffic and throttling just the device/ip it was coming from. But now I think it must be the device’s network stack itself (or maybe interaction with the wifi/modem) that’s getting hung up, as the response time is too fast as soon as Last Epoch is no longer doing its thing. It can’t be my ISP connection in general since other devices on the same wifi keep working fine.

Tried running wireshark and then network analyzer. For TCP I just see some retries happening, but not that many packets. For UDP, almost all of my communication was happening with a server at 216.223.24.98 which must be the game server. Usually my machine is sending 100-200 UDP packets per second, but during the lag spike this went up to over 2000 packets per second, hung around 1500 pps for about 20-30 seconds, and then subsided to 100 pps. During this time my ping to google went to 2s, 3s, and even timed out for a few iterations.

Over my ~1 minute of packet capture I sent out 41k UDP packets to the game server and received 100k. I didn’t examine the download side, kind of assuming that’s not the issue, but that needs sanity checking.

Analyzing the packets themselves the payloads are generally pretty tiny. I can identify some Binary FrameStream objects (CONTROL_BYTE=130, payloadLengths of 40, 46, 58, 65) with various RouteIds like CREATED_OBJECT_ROUTER_ID and RPC_ROUTER_ID.

However the majority of the packets I sent have a payloadLength=17 and the first byte doesn’t seem to be a CONTROL_BYTE. The first byte value is all over the place in these packets, but the values are varying only slightly from packet to packet. But I’m not sure of their source as they don’t seem to follow the FrameStream format. Sample payloads here look like:

8D 7A 07 00 00 00 00 00 1D 39 00 00 00 00 00 00 37
92 7A 07 00 00 00 00 00 1E 39 00 00 00 00 00 00 37
95 7A 07 00 00 00 00 00 B3 38 00 00 00 00 00 00 37

Notably, I don’t see any Ping packets (CONTROL_BYTE=132), so not certain this is the bug referenced above. I saved the packet dumps (.cap and .matp) if someone from LE wants to take a look.

1 Like

Sweet! I’d say that’s actually good news. Then maybe they should look into Forge Alloy as a quick and painless replacement. Another thread that should probably be associated with this. They closed it already without communication, still very much an issue.
https://forum.lastepoch.com/t/connecting-to-online-characters-bugged/54312/24

God the desync and having to relog constantly is so frustrating in this game hopefully next patch they address it.

2 Likes

My friend started having these issues recently and keeps needing to relog every other monolith to be able to play them (and happens often in the campaign as well).

Well i give up, the lack of acknowledgment to this issue is heart breaking.

Hope it gets fixed in 1.0

1 Like

Reported data loss via the support page.

New to the game and had been enjoying the game until now.

Lost from Chapter 8 to Chapter 9. Was about level 54 and it backtracked me to 49. What are the chances they are going to be able to restore this data? Does playing offline save the data on the PC/Steam rather than with EHG? Can we move the save data from online to offline?

It’s just unreal to play in online version on glasscanon dynamic class, too much desync/lags of all things on the screen (projectiles, aoe, character’s abilities and monster’s casts)… Hope this network stuff is high-priority for developers

I’m pretty new, bought the game after 0.9 came out, and have been having all sorts of desync problems, especially during peak hours. Excited enough about the game to come here and +1 this post… it’s had me wanting to ragequit several times in my first ~9 hours of gameplay.

really it’s a testament to the quality of gameplay that I’m here posting at all… please please please fix this.

Multiplayer is obviously not ready for release. This is ridiculous - I’ve had no response from support on what’s been done to work on this issue, and the game is unplayable in multiplayer.

Maybe EHG should test things properly before releasing them. Also it would be a good idea if they communicated properly with people to let them know what the issue is with this, and what’s been done to fix it.

I wouldn’t recommend anyone buy this game at the moment with it in such a poor state, which is a pity, because it was fine up until multiplayer was released.

Hi All,

I don’t know if this is coincidence or not - I checked my modem tonight, and it has UDP flood protection. Default was 2000 packets/sec with a 10sec timeout.

I changed it to 8000 packets/sec and played for an hour straight afterwards. I did notice some very brief pausing that lasted less than a second at times.

I don’t know if this has changed anything or if something else has changed - I’ll test more tomorrow.

After 0.9g, zones still randomly take extremely long to load. Sometimes a zone will load almost instantly, sometimes the same zone will take 5+ mins to load.

Campaign is a pain to do because zones are so small that I spend more time loading than actually doing stuff, and quests like A Study in Time that require 6 zone loads minimum are an absolute joke, 4 mins of gameplay for up to 30mins of loading

Im still running into infinite loading screens and desync. heres hoping sooner or later they fix it.
I recommend everyone learn how to record clips with Nvidia Geforce Experience and similar software to send to EHG support team so they hunt down the bugs.
its all we can really do to help them figure it out to be fixed.

Game is still in beta and not released. I actually have had no issues in multiplayer with my friends. Other end of the coin here.

You have an apparently significant part of the playerbase dealing with a problem that clearly affects playability, and your answer is basically: “I have no issues, so it’s not a problem”.
That is not adding anything constructive to a conversation.

Dealing with the same issue, if the devs publicly acknowledge it and request information, I will gladly provide what I can.

1 Like

Optimization part of the patch notes for 0.9g does reference these issues, so I do think they acknowledge the issue.

Except if you were active on areas such as discord, you would see them live responding to people and acknowledging this exact issue. It’s easy to get frustrated and blinded because you can’t play and lash out. Try and do a bit of research before sitting on the forums accusing the devs of ignoring the issue :). As per your comment about “apparently significant part of the playerbase”… People who cannot play will always be the loudest on the bug forums, you have no internal metric numbers indicating what percentage of the playerbase is affected so that claim is unproven and not based in fact.