New network going up now, testing 4MB chunks. Time to reset your nodes

Development Updates
1

Hi everyone!

  • The new network is going up now, time to reset your nodes.
  • 32 GB Nodes with 4 MB chunks.
  • You should be on Safenode 0.111.4

For Node Launchpad Users:

  1. Open the Node Launchpad.
  2. Press O to access the Options screen.
  3. Then, press Ctrl R, type reset, and hit enter. This will stop your nodes, reset, and upgrade their software with the latest network updates.
  4. If you haven’t added your Discord Username yet, press Ctrl B to do so.
  5. Press Ctrl G to restart your nodes.
  6. Once your nodes are running, use the /rank command here on Discord to get on the leaderboard, and check your progress.

For CLI Tool Users:
If you’re using the CLI tool, you’ll need to reset your nodes as well, using the latest Node Manager (v0.10.6).

The full reset command guide is available on our documentation site.

A few other notes:

  • This network will run for a week (Thurs-Thurs). The last release before launch will be 2 days shorter then expected with this change. Both the launch date and overall phase 2 period of the beta have NOT changed.
  • The /rank won’t include any nanos earned on the Oct 8 network, just the Thursday–Thursday. We’ll add on those extra nanos to the leaderboard published at the end of the week
  • Remember to ramp up your nodes gradually, expect increased RAM/CPU usage compared to previous networks with smaller chunk sizes. You can anticipate a 2-3x in RAM usage compared to what you were seeing with smaller chunk sizes prior to Oct 8. The exact resource use isn’t confirmed as that is part of what we intend to learn from this test!
  • We will be ramping uploads gradually
  • Devblog will be out today going into the events of Oct 8
12 Likes
3

Initial observations

  • seeing similar (not as bad) b/w uploading from my nodes and will be due to new nodes and they are retrieving chunks from the nodes.
    • not as bad because there is only around 50-100 chunks being stored on the nodes. basically empty network to coddle the network and hope it doesn’t get a transmission storm like the last one
  • Seeing a very high figure for each node seeing other nodes are bad. Obviously there are many nodes out there not doing as well as mine
  • no node is being shunned yet by others

20 nodes so far

3 Likes
4

keep reporting progress please :star_struck:

2 Likes
5

I’ve noticed that most of the records in record_store are actually 512KB ones. There are only a few 4.1MB ones so far.

safe@sn-test-02:~ $ ls ~/.local/share/safe/node/safenode1/record_store/ | wc -l
520
safe@sn-test-02:~ $ ls -lh ~/.local/share/safe/node/safenode1/record_store/ | grep 513K | wc -l
451
safe@sn-test-02:~ $ ls -lh ~/.local/share/safe/node/safenode1/record_store/ | grep 4.1M | wc -l
10
2 Likes
6

only takes a few when all are being uploaded to satisfy news nodes joining at a large rate

3 Likes
7

did someone upgrade and bring in “old chunks” ? I didn’t expect this to happen in the current network :open_mouth: … since chunk size is larger … but we had that multiple times in the past … so I guess it might have happened again

3 Likes
8

I’m guessing Maidsafe have a testdata workflow which still produces chunks of 512KB and that is running. Either by accident or design. If it’s by design maybe it’s so that the initial data in the network is more easily replicable stuff.

1 Like
9

No the team said they did uploading prior to releasing the contacts file. That is where the records came from

3 Likes
10

…we had data transitioning from every singled network to the next one so far …

and the previous chunks were exactly that size … why would the pre-upload smaller chunks? …

3 Likes
11

Because nearly every file will have the last chunk a small one and all file < 12 MB will have smaller chunks.

1 Like
12

@storage_guy has 450x 512KB and 10x 4MB

1 Like
13

Did he reset his nodes? If not then he has old chunks that the node will ignore now.

All the new nodes have less than 200 chunks MOST less than 100 records. I have none over 100 records stored.

1 Like
14

just checked a random node - 2/3 of chunks are 512KB - and yes I did a full nuke in preparation of the update

1 Like
15

I just went and checked mine and they are now in the hundreds. One I checked has over 500 records and 400MB

16

that’s data expanding from nodes that upgraded instead of reset and the network doesn’t prevent that old data from entering the network

@joshuef @qi_ma is this expected behavior?

2 Likes
17

Seems some people have just been upgrading their nodes and the records in their record store is being given back to the network. They have not been resetting.

For instance safenode-manager uses the same directory names so if the directories are not deleted then the record store remains. No reset then the directories are not deleted

People have pointed this out every beta network since the first time it was noticed. Some of these chunks are likely to be from a few betanets ago

5 Likes
18

Yes, total reset.

1 Like
19

Possibly some from the first beta network that ever existed :sweat_smile:

4 Likes
20
8 Likes