5

I have been running a node for about 6 months now and my chain data recently maxed out my ssd completely and I started seeing disk I/O errors in the node's log output.

I made a mistake snapshotting, etc...and am rebuilding the node from scratch — it's currently bootstrapping.

With the new code and new node, after the bootstrapping with the chain data take under 16GB, or will it still take 150GB-ish, I'll I need to snapshot and upgrade it again?

Update

I should add I'm using Docker and the mainnet.sh script. I created a fresh node docker container (I deleted the VPS, recreated it) with the latest image and let it bootstrap. After bootstrapping I saw, I double-checked which mode I was running in:

$ ./mainnet.sh client rpc get /chains/main/checkpoint

{ "block":
    { "level": 442369, "proto": 3,
      "predecessor": "BLVxE1WGhVhtyFfYLDdKhe42N25nTGYWjKvfA2AokHDM6vJMjSQ",
      "timestamp": "2019-05-18T06:39:35Z", "validation_pass": 4,
      "operations_hash":
        "LLoazsHZS8KSdCNFe9B6u5y78nYCynni8tqGEACKPywmrkisE1FEz",
      "fitness": [ "00", "0000000000d35c7d" ],
      "context": "CoVCLRtu1TeX8s5BBrQ3db56epGZCx7DH6vHS6Tw7QBFzbkV3HEM",
      "protocol_data":
        "00000000000373f6f25a004b67f5443ea32df421c4dbb31b3d14bbb2b297587261e2214c69adc43157040d9578272326ea60d97eb5f01706866923f7c2b88ffe2b4dbe7e7e5901ae450300" },
  "save_point": 442369, "caboose": 0, "history_mode": "full" }

I checked the disk space after bootstrapping completed:

5.0G    store
113.9G  context

Then per the docs here, I upgraded with:

./mainnet.sh upgrade

Now checking the disk space a day or two later or I see:

5.8G    store
140.7G  context

What is wrong here?

Is it also strange that rpc get /chains/main/checkpoint outputs a proto of 3 instead of 4?

Improve this question
4

On the new node in full history mode my bootstrapped folder looks like this:

store 5.5 GB 

context 13 GB

If you updated your node and started rebuilding the blockchain from scratch your node should be running in full history mode by default. There is no need to snapshot and upgrade it again.

Improve this answer
8
  • Thanks. Things are still bootstrapping (at Feb 2019 now), and the disk space is growing at a rate in which it will run out. Currently 54.6% of 154.90GB is being used and by the time I get to 'today' the disk will be full. After the bootstrapping is complete is there some cleanup of store and context that takes place?
    – lostdorje
    Jun 1 '19 at 5:55
  • Not sure about the clean-up after bootstrapping. I suggest you import from snapshot (e.g. from here: tzdutch.com/quicksync). Make sure to delete your 'store' and 'context' folders in your .tezos-node directory before. The download file is 1.5GB-ish. Once you extracted and imported it it will be closer to 18GB but at no point in the process much higher than that. Instructions for snapshot import: tezos.gitlab.io/master/releases/may-2019.html
    – Andreas Tissen
    Jun 1 '19 at 19:51
  • I'm using docker a brand new container I just pulled/built (also in full mode, I double checked). Then I ran 'mainnet.sh upgrade' per the docs and my store is 5.9GB and context is 140.7G. I wonder if the docker upgrade isn't working.
    – lostdorje
    Jun 3 '19 at 8:24
  • You don't need to run mainnet.sh ugrade if you are already on the newest software and have no old blockchain file. Instructions following in the next comment.
    – Andreas Tissen
    Jun 3 '19 at 11:00
  • 1
    Do the following: 1. clean install a node on your machine (doesn't matter if docker or anything else; make sure your .tezos-node folder doesn't contain store and context folders) 2. download a snapshot from tzdutch.com/quicksync and unzip it. Instructions are in the link. 3. Navigate to unzipped file and execute command: tezos-node snapshot import mainnet-*.full 4. Start node with tezos-node run. Now you can execute tezos-client bootstrapped in a different shell and should see that the node is roughly a week behind. It will typically sync within a couple of minutes.
    – Andreas Tissen
    Jun 3 '19 at 11:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.