Mina's Adversarial Testnet, Testworld, is here!
Check it out



Testnet Status:



Here are some common problems you might encounter while trying to set up the Mina daemon. If you can't find your issue here please ask for help on Discord or open an issue on Github.


My node crashes every few minutes?

If the node crashes quickly and repeatably, this is likely a configuration issue such as incorrect permissions on the private key, failing to find peers due to incorrectly passing the peers.txt file, an incorrect password or special characters in the password. The last few lines of the logs should provide more information on the issue.

I closed my SSH terminal, and the node crashed?

If you start the mina daemon in the foreground, once you detach from it, it will shut down (via a SIGINT signal). To run the process in the background so that you may detach, you have many options:

  • Start the daemon with the -background flag.
  • Run as a systemd service (see the docs).
  • Run with Docker (see the docs).
  • Run the daemon in a terminal multiplexer such as screen or tmux that you can detach from to keep the process running.

Running as a service / via Docker is recommended as they have the advantage of automated restarts if your node crashes.

What permissions are required for the keys directory?

The keys directory should have 700 permissions, and the private key file 600. For example, these commands update the permissions when the keys directory is contained within your home directory and your private key file is named my-wallet:

chmod 700 ~/keys
chmod 600 ~/keys/my-wallet

Can I run on a Raspberry Pi / ARM-based device?

No. Only x86-64 is supported and not ARM.

I have special characters in my password, and it is giving an error?

You should use quotes around the password i.e. "MY_PASSWORD" and/or you can escape the special characters such as $ in the password using the \ character.

I get a permissions error when running the docker command?

Add the current user to the docker group as per this article. You could prefix the command with sudo, but this is not recommended.

I see monitor.ml.Error "Timed out getting connection from process" on macOS?

If you're running Mina on macOS and see the following time out error monitor.ml.Error "Timed out getting connection from process", you'll need to add your hostname to /etc/hosts by running the following:

  • $ hostname to get your hostname
  • Open the /etc/hosts file and add the mapping:
# Host Database
# localhost is used to configure the loopback interface
# when the system is booting.  Do not change this entry.
##       localhost       <ADD YOUR HOSTNAME HERE>
This is necessary because sometimes macOS doesn't resolve your hostname to your local IP address.

What hardware do I need?

See the requirements for the current testnet here. Some lower-tier VPS providers may have an issue producing blocks within the slot time. If you continue to experience missing blocks, you may wish to try on more powerful hardware.

I would like the mina service to start on boot?

If you use systemd to manage the Mina daemon then it will not automatically restart on a reboot of the machine. You can manually run the service via systemctl --user start mina or run systemctl --user enable mina to have the service automatically start on boot.

What should I add for CODA_PRIVKEY_PASS?

When you generated a keypair and submitted it to participate in the testnet, you created a password. Use the password you created for the value of CODA_PRIVKEY_PASS.

I get an error about ~/.coda-config/daemon.json missing?

This is an optional file to pass configuration options to the daemon (rather than passing via the command line). The daemon looks for it by default, which results in an error if you do not include it. Just ignore this error (unless you are trying to include it) as it is not fatal, and the daemon will start as per normal.

I see messages about prover/verifier being killed periodically

This is normal, and these processes are currently periodically killed and restarted. Ignore these messages unless they result in a fatal error/crash.

I crashed with "Fatal error: out of memory"?

This is a common issue - see this issue on GitHub and add any relevant information with logs and restart your node.

Can I provide my password using an environment variable?

Yes. Set the CODA_PRIVKEY_PASS environment variable to be your password.

I am trying to install on Ubuntu 20.04 / Debian 10 and get dependency errors?

See this guide to manually resolve the required dependencies. Alternatively, use a supported OS (Ubuntu 18.04/Debian 9) or use Docker.

Syncing a node

I am stuck at block height of 1 and a max observed block length of 1

The output of coda client status outputs similar to the following:

Block height:                    1
Max observed block length:       1

This issue was fixed by the release 9652f8ee092ea77e29f5ab49fa0a295e36743e8b, please ensure you have upgraded. It is normal to see a height of 1 while the daemon initializes. It could take several minutes (20 mins to an hour) before the node progress to the catchup phase.

My "Max observed block length" is 1 but my block height is correct?

This issue was fixed by the release 9652f8ee092ea77e29f5ab49fa0a295e36743e8b, please ensure you have upgraded. After upgrading make sure you let the daemon run (20 mins to an hour), and the error should resolve itself.

My node is stuck in catchup / the block height does not update

When the node is synced, Block height should be equal to Max observed block length, as seen in the output of coda client status and should match the current height of the network, as shown in the screenshot below:

Syncing a node

When starting the node, and the node is in a bootstrap/catchup state, the block height will display 1 or a number that is a few hundred blocks from the current best block length (but much greater than 1). During catchup, the block height will not incrementally update, other than when new blocks are produced and will jump to the correct height when catchup is complete. So be patient; this process can take several hours to complete and does vary depending on the hardware used.

I synced at height 1 / not at the current height of the network?

What matters is that your current blockchain length, as seen in coda client status, is equal to the current network height. If your node is reporting it is synced at the wrong height, it should eventually enter the catchup phase to get the node to the correct height. You may occasionally switch between catchup and synced if the network sees any short forks it tries to catch up to.

Why does my blockchain height jump to 1000+ and then get stuck?

We only need the last k (290) blocks of the blockchain to produce blocks. So once bootstrap is completed, the root of the transition frontier (essentially your local store of blocks) should be the current maximum height minus k (290). The part of the syncing process that takes significant time is downloading those missing k blocks, which is the role of catchup. During catchup, you won't see your block height increase (other than new blocks produced on the network) until it jumps to the current height and is synced. This takes time, so leave it running!

My sync status is offline?

This status indicates that you have not received any messages from peers for the last ~24 minutes. Ensure that you are using the latest peers.txt file. You can add new peers to a running node via this command:

cat peers.txt | xargs -I % coda advanced add-peers %


My hosting provider is warning me: "Abuse message: Netscan detected"

There is some behavior with our p2p networking module that triggers this warning on some hosting providers (specifically we've heard this happening with Hetzner). To mitigate this configure your firewall to allow traffic from ssh, http, https and deny outgoing traffic to all private ip addresses.

Here is an example of this using the ufw firewall tool. Thanks Ducca for sharing these rules and confirming they fix the issues on Hetzner.

Allow SSH, HTTP, HTTPS porst:

ufw allow 22
ufw allow 80
ufw allow 443
ufw enable

Block outgoing private connection:

ufw deny out from any to
ufw deny out from any to
ufw deny out from any to
ufw deny out from any to
ufw deny out from any to
ufw deny out from any to

See #7053 for more

What ports need to be open?

The only port required to be open for the daemon to communicate with peers is TCP port 8302.

The client port 8301 should never be exposed to the internet. There may be outdated references to other ports that were once used before but are no longer required.

Node fails with "Failed to find any peers, crashing as this is not a seed node"?

First, ensure that you have downloaded the latest peers file from the email received - the URL does not change, but the list of peers is updated very frequently.

Next, depending on the method you use to start the node, ensure that you are passing the correct path to this file. For example, if using systemd, there must be a peers.txt file in your home directory.

This error should have been resolved by a recent update, so ensure you are on the latest release and have the latest peers file.

I get an error couldn't determine our IP from the internet?

If you see couldn't determine our IP from the internet, use -external-ip flag, then the daemon failed to determine its own IP from these service providers. Your firewall may be blocking HTTP/S requests, or you have a misconfigured network connection.

To bypass this, pass in the flag -external-ip <your-ip-address> when starting the Mina daemon. To get your external IP address, run curl ifconfig.me.

Do I need to configure port forwarding manually?

The Mina node does its best to configure itself to be able to connect to the outside world without you needing to do any extra work. However, this may fail, depending on your router and network setup. In this case, you may have to manually forward the external-port.

A common cause of this is routers not supporting UPnP, a protocol that allows the node to configure the port forwarding automatically. If you experience this type of problem, find your router model and search for <model> port forwarding and follow the instructions to forward the ports from your router to your device running the Mina node. You'll need to open the TCP port 8302 by default.

Note: When running Mina in the cloud, you should instead configure security groups for your cloud provider.

Do I need to allow "Accept incoming connections"

If you see one or more warnings like the below, then choose "Allow":

Do you want the application "mina" to accept incoming network connections?

Accounts and Transactions

What fee should I use to send a transaction / my transaction is stuck as pending?

Transactions that are included in blocks are prioritized by their transaction fees. If many transactions are pending, then the highest fee transactions will be included first. So, increasing your fees is the best way to ensure that your transaction is included. To view the transaction pool on a running node, you can use the following command (you will need to install jq, which is used to format the output of the following command, e.g., sudo apt install jq):

coda advanced pooled-user-commands | jq . | grep fee | sort | uniq -c | sort -n
      1     "fee": "0.031",
      5     "fee": "0.1",

If there are many transactions in the pool, you may need to increase your fee. For a visual overview of transaction fees, see here.

If you have a stuck transaction due to a low fee, you will have to wait for it to be included in a block or to be evicted from the transaction pool as there is currently no way to cancel a transaction due to this known issue.

How many transactions can be included in a block?

On the current testnet, this is 128, which also includes the coinbase transaction as well as any fee transfers to pay snark workers.

Can I cancel a transaction?

Not currently due to this bug.

I sent a transaction, but it was never included in a block?

See this answer. The transaction was likely stuck as pending and then eventually evicted from the transaction pool. Try sending again.

Why is the coinbase 0 / there are no transactions included in a block?

In certain circumstances, there is no snark work available to be purchased or is too expensive to be purchased. In this case, it may not be possible to include a coinbase transaction, and so there is no coinbase awarded for the block.

If transactions are not being included, the transaction fees likely do not cover the cost of the snark work required to be purchased to offset, including the transactions.

Does the order of sending transactions matter?

Yes, transactions are processed in order according to the nonce associated with the transaction. So, if you send a transaction that is stuck, with, for example, a low fee, all following transactions will also be stuck regardless of the fees used in the later transactions. Unfortunately, there is no way to currently cancel a transaction due to a known issue, so you have to wait for either the transaction to be included or to be evicted from the transaction pool.

Why does my account say "locked"?

If the output of coda accounts list shows your account is locked, this means that you need to unlock it using your private key password in order to use it to send transactions. To unlock, simply use the coda accounts unlock --public-key command.

A block producer does not need the account to be unlocked to produce blocks.

I requested funds from the faucet, and they never arrived?

Check the status of the faucet here to see if there are any know issues with the faucet.

You can also check this explorer to see all pending transactions from the faucet to see if yours is listed or if there have been any recent faucet transactions.

If you have just started your node and are not synced, any balance sent from the faucet will not be displayed until the node is synced.

What are time-locked accounts / why are some accounts timed and others untimed?

See the docs here.

Can I send a payment before I am synced?

Yes, but only if you have funds already in the ledger. If you requested funds from, e.g., the faucet or another community member, you will need to wait until you are synced so that you have a balance you can use to send.

Error: Specified sender is not in the ledger or sent a transaction in transaction pool when sending a transaction

See this answer, you will need to wait until you are synced if you do not have an existing balance in the ledger.

I am running a block producer, but I don't see anything in coda accounts list?

If you started the daemon passing the block-producer-key flag, then you still need to import the account to the daemon in order to send a transaction. To do so, use the coda accounts import -priv-key command, passing in the location of the private key file.

Block producer

How do I run a block producer?

The methods to start a node in the documentation will run a block producer by default.

How do I know if I am running a block producer

To check the status of your block producer, run coda client status and look for the lines Block producers running. You should see a value of 1 and also your public key.

Running a block producer

The line Next block will be produced in lets you know the next time you have won a slot to produce a block. If you have not won any further blocks in the epoch, a message "None this epoch..." will be displayed and will only update once a new epoch starts.

How can I increase my chance of winning a block?

The chance of winning a block is determined by a Verifiable Random Function (VRF), with the chance of winning a block being proportional to your stake. The VRF will always return the same result no matter how many times it is run, so there is no way of improving your chance of winning a block. Other than your staking balance, it is down to luck.

The stake associated with your public key is determined in advance at the start of an epoch. There is a delay in it being considered, so receiving funds, block rewards, delegating other funds, and snark worker rewards will not improve your chance of winning a block on this testnet as these funds will not be considered in the staking ledger for at least another full epoch (~2 weeks) by which time the testnet will (likely) be over.

Why was my block orphaned / why did I not get a block reward?

There can be more than one block producer per slot due to the way block producers are selected. So if two (or more) blocks are produced for the same slot, it will cause a short fork, and only one will be chosen - the winner in this instance is random based on the VRF output for each block producer. If you consistently see orphaned blocks, then you may be producing blocks slowly, and so other block producers may not see your block before building atop the current longest chain.

It is also possible to produce a block in catchup, and this block will also be quickly orphaned as it is not building on the correct height.

I didn't produce a block / I missed a slot?

There are a few reasons why you could miss a slot / not successfully produce a block, for example, the node restarting and being in bootstrap at the time of producing a slot.

Also, you must produce a block within the slot time (3 mins). If you are on less powerful hardware or the daemon is competing for resources, it may not produce the block in time. In this instance, you should find in your logs:

Internally generated block $state_hash cannot be rebroadcast because it's not a valid time to do so ($timing)

It is not recommended to run a snark worker on the same machine at the same time you are producing a block as the snark worker is resource-intensive and may lead to the block not being produced in time. You can disable the snark worker during block production by, for example, using the snark stopper script.

Why is the block rate so low / how often should there be a block?

A slot on the current testnet is every 3 mins, though not all slots should have a block produced, so on average, we would expect a block every 4 mins. However, not all the stake is online and active in producing blocks, and so not all slots will have a block, and sometimes there can be long delays between blocks. As more of the stake is online and staking, this situation improves.

Why does O(1) Labs win most of the blocks?

To keep the network stable, O(1) Labs has 30% of the stake, so has a much greater probability of winning a slot and producing a block. Users who were given a balance of 66,000 mina have about a 0.1% chance of winning any particular slot, so should expect a block every couple of days, but luck plays a major factor in how often you will win a block.

How do I know if I won a block or a transaction went through?

The easiest way is to check a block explorer. You can also get this information for recent blocks via the GraphQL API.

Why do I have a message of "No blocks won this this epoch"?

See this answer.

Are orphaned blocks considered for the challenges?

Yes, during this testnet, orphaned blocks will be considered.

Snark worker

My snark work is not getting bought / what fee should I use for a snark worker?

If you are running a snark worker and not seeing any work being included in blocks, then likely others are producing cheaper snark work. Multiple snark workers are all competing for the same snark work, with only the lowest fee for each being included in the snark pool to be bought by block producers.

Sometimes high fees will be included in blocks, and this is a function of how snark workers select which work to complete and which work is required to be purchased by the block producer.

By default, the work selection for a snark worker is random. You can change this by adding the -work-selection seq flag to the coda daemon command, which will work on jobs in the order required to be included from the scan state and will likely result in your snarks being included without a potentially lengthy delay.

For choosing fees, you can look at historical blocks to determine prices that have been bought, or look at the table listed here.

How can I disable the snark worker?

Run coda client set-snark-worker to disable the snark worker. To enable again, pass your public key coda client set-snark-worker -address <YOUR_PUBLIC_KEY>.

Can I run a snark worker and block producer on the same machine?

Yes, you can, but you should stop the snark worker during block production so as not to compete for resources and miss producing a block. Snark workers also consume more resources in general. See the snark stopper script to help in automating this.


I see weird messages/errors in the logs?

The logs are noisy and often contain "scary" looking messages such as failure to connect etc. As a general rule, if the message is not fatal and the node does not crash, it is likely nothing to be worried about. The below messages are all considered "normal":

"RPC #841 failed: \"internal RPC error error: unknown stream_idx\""
"Peer $peer didn't have enough information to answer ledger_hash query. See error for more details: $error"
"Timed out waiting for the parent of $cached_transition after 0 ms, signalling a catchup job"
"Failed to reset stream (this means it was probably closed successfully): $error
        error: {
    "RPC #365 failed: \"internal RPC error error: unknown stream_idx\""

How can I get my logs running as a service?

journalctl --user -u mina -n 1000 -f

How do I get my logs running Docker?

docker logs --follow mina

How do I get my logs running the daemon manually?

tail -f ~/.coda-config/coda.log

How many logs are kept / where are the logs located?

The ~/.coda-config directory contains the Mina logs. This directory contains the following:

  • coda.log - This file contains the latest logs of the daemon. Each log file is limited to 10 MiB in size and rotates through 50 log files. Rotated log files are named coda.log.x from coda.log.0 to coda.log.50.
  • mina-best-tip.log - This is used to write the best tip logs to make it easier to collect the required logs from nodes to determine the state for a hard fork. Each file is limited to 5 MiB and rotates through a maximum of 5 files from mina-best-tip.log.0 to mina-best-tip.log.5.
  • coda-prover.log - This logs memory usage and batch size of the prover and is limited to 128 MiB and rotates via a single log file.
  • coda-verifier.log - This logs memory usage and batch size of the verifier and is limited to 128 MiB and rotates via a single log file.
  • coda.version - This file contains the Git SHA of the running daemon.

Testnet Challenges

I lost the password for my private key I submitted for the testnet?

If you lost the password for the private key, then there is nothing that can be done to recover it. If you were selected for the testnet, the ledger may have a balance associated with this public key, and it isn't possible to move this to a new key. You can still participate, however, by creating a new keypair and getting a node running and synced. You won't have a balance on the key, so you will not win any blocks but can participate in running a node and in the other community challenges such as running a snark worker.

Always keep a copy of your private key you generated (likely called my-wallet in a folder named keys) and the associated password!

I completed a challenge, and the leaderboard has not been updated?

As per the message in the #leaderboard-qa channel on Discord, please be patient!

We're still working on a fix for the leaderboard. We have recovered the block data for the service that went down, and we're in the process of restoring the Leaderboard service. You can expect some updates soon.

I was on the waitlist, and now my balance displays 0?

If you do not have a balance for your private key, check that you are using the same key that you used to register for the testnet.

Not all waitlist participants have a balance, and if you are one of them, you can still sync a node and request funds from the faucet or other community members. Your node will need to be synced before any balance from the faucet or community members will be visible via coda accounts list. There may be additional stake redistributions at a later time, assuming you have a synced node.

Do I need to send 50 tx to different addresses / to what address should I send?

You can send the transactions to any address. It doesn't matter what fees/amounts/addresses you use - just 50 transactions from your public key. The key has to be the one registered for the testnet as it is associated with your Discord ID.

Is there a guide to completing the testnet challenges?

See this guide by community member @Ducca.

Where can I find a list of all the current testnet challenges?

See the post here.