Releases: syscoin/syscoin
Syscoin v5.0.4
This release will allow nodes to sync passed block 2032320. If your node is stuck on this block you may upgrade to sync. You should reindex upon upgrade.
Full Changelog: v5.0.3...v5.0.4
Syscoin v5.0.3
This is a non-mandatory update that focuses on optimizing memory and disk usage of sentry node lists and PoDA (Bitcoin DA). Please reindex or resync your node upon upgrading to 5.0.3. Miners should update due to the increased minimum gas price (tips to pay to miners) and increased gas limit signalled in the block. Due to the fixes to the storage and disk usage it is highly recommended for nodes to upgrade.
Gas defaults and miner tip increase
We increase the miner tip for the eip1559 transaction type up 100x from current levels. This will filter out transactions paying little to no tips to miners. We also increased the default gas from 8m to 16m to support larger deployments on NEVM. Miners should update to signal support for the increase.
Chain parameter update
Updated checkpoints, noticed on testnet nodes were having issues pre-syncing headers passed the nexus block.
Reworked Proof-Of-Data-Availability (PoDA)
Memory management
Used shared pointers for safer memory management and moved blobs into a seperate database to optimize cache and memory load only when needed, use a metadata database for fetching the stats of blobs (for example when checking if exists, or when to prune)
Proper serialization of transactions with PoDA
Previously we weren't able to deserialize and display blob transactions as they were not provided properly when getting the raw transaction via getrawtransaction, and when trying to decode via decoderawtransaction it would fail since the blob data is stripped out normally. Added a serialization flag to be able to decode the transaction properly (still without PoDA - just the version hash exists in the OP_RETURN). SER_NO_PODA is passed into all raw tx APIs (REST, RPC).
Fix ScanBlobs RPC
Rework scanblobs to reflect the new way to store and fetch blobs, and accurately return stats for all blobs in cache/db.
Sentry node list (evodb_dmn) memory/disk size improvement
Flush only when full cache
Previously the cache was waiting to be full before the leveldb folder is recreated ensuring the last 1728 lists are stored to optimize the size to around 1.4gb (for 1728 lists of around 2800 sentry nodes on mainnet for each list). However the cache was never allowed to get full because on every flush (once an hour or on shutdown) it would clear cache and write to the disk preventing the full cache logic to recreate the leveldb disk folder. Now we prevent the flush (once an hour or hard flush once a day) from writing the cache to disk and keep it in memory, unless the cache is full then we write to disk by recreating the db.
Recreate DB via CDBWrapper::ResetDB()
Created a new helper to delete/recreate the evodb database underneath rather than recreating the entire evodb object which loses is cache (needed to copy the cache contents which spiked memory on flushes for the evodb_dmn list flush). Previously we were storing copies, recreating DB, resetting copies and then flushing to disk. This created 2-3gb memory spikes because of the large size of the cache of 1.4gb+, once for copying cache to memory and once when writing on the flush in one batch. We created a helper to delete the DB without affecting evodb object so the cache can remain in place and flush to disk as usual.
Flush in coordinated batches rather than one big batch
When we flush in one big batch its allocating a copy of the data to batch before writing. We instead write in batch segment while clearing cache for the written segment to only allow one segment to be additionally allocated at a time. For evodb_dmn for example we use a segment size of 256 so 256 entries are written per batch call, then a new batch is started subsequently until up to 1728 are stored. This doesn't seem to affect performance (especially on SSD) but greatly minimizes memory overhead and spikes. On flush now I notice previously 1-2gb memory jump reduced down to 150-300mb overall across all of the db flushes. We apply this logic to a few other database flushes as well trading off a bit of speed for writes with lower memory footprint during the flush.
Various upstream merges
Merges from upstream on sysgeth and syscoin for quality-of-life improvements and optimizations.
Full Changelog: v5.0.2...v5.0.3
Syscoin v5.0.2
This is a non-mandatory Hotfix update focusing on networking synchronization and testnet fixes related to Sentry nodes. All Sentry nodes must update on testnet and mainnet.
Full Changelog: v5.0.1...v5.0.2
Syscoin v5.0.1
This is a non-mandatory Hotfix update focusing on:
- Legacy BLS synchronization: Pulled in the latest BLS library and migrated the codepaths depending on the changes from that library.
- Quorum establishment: Reworked networking code to better align with latest Bitcoin core.
Full Changelog: v5.0.0...v5.0.1
Syscoin v5.0.0 (Nexus) Release
Syscoin 5.0.0
This is a MANDATORY upgrade for all nodes.
Release notes https://github.com/syscoin/syscoin/blob/master/doc/release-notes/release-notes-5.0.0.md
Official block height for Nexus is on block 2010345
this is an explorer https://explorer-blockbook.syscoin.org/ you can follow for Syscoin blocks.
Nodes can upgrade prior to the block. You may install by simply running the new version of code overtop of their existing install. Reindex should happen automatically which will reindex the chain from disk. If it does not, you should either reindex or do a full sync again.
NEVM explorer: https://explorer.syscoin.org/
Ethstats: https://ethstats.syscoin.org/
Sentry nodes: You do not need to run sentinel anymore, its built-in to the node.
Upgrade guide: https://syscoin.readme.io/v5.0.0/docs/syscoin-50-upgrade-guide
Full Changelog: v4.4.2...v5.0.0
Syscoin v5.0.99 (Nexus) Release - Testnet
Syscoin 5.0.0 (for testnet not for mainnet yet)
This is a MANDATORY upgrade for all nodes.
Release notes https://github.com/syscoin/syscoin/blob/master/doc/release-notes/release-notes-5.0.0.md
Official block height for NEVM fork is on block XXXX
this is an explorer https://explorer-blockbook.syscoin.org/ you can follow for Syscoin blocks.
Nodes can upgrade prior to the block. You may install by simply running the new version of code overtop of their existing install. Reindex should happen automatically which will reindex the chain from disk.
If one does not want to run Geth for NEVM part of the network you can define zmqpubnevm as an empty string which will skip running Geth and not validate NEVM blocks.
NEVM explorer: https://explorer.syscoin.org/
Ethstats: https://ethstats.syscoin.org/
Sentry nodes: You do not need to run sentinel anymore, its built-in to the node.
Full Changelog: v4.4.2...v5.0.99
Syscoin v4.4.2 Release
Hot Fix for the following issues:
- coincontrol issues not selecting the right amounts and crashing if using the "use available balance" button
- presync looping (i think was geth in invalid state, so we just disable geth connection for the sync)
- sendrawtransaction(for sys burn to nevm we skip the max burn check). This should fix pali/blockbook
- gobject_prepare fix which wasn't signing before sending the transaction. Useful for people creating governance proposals
- Large rollback on nevm fix, which causes general issues related to pruning PoDA when transaction is removed from mempool as a "reorg", we shouldn't do this because upon subsequent block validation the block will not validate.
Syscoin v4.4.1 Release
Hot fixes for users of Syscoin Core on OSX ARM M1/M2 and Windows:
- non-supported geth systems like arm OSX M1/M2 should not run geth at all and shouldn't stop at blocks that require NEVM
- prune was being enabled on some systems by default which disabled governance which led to coinbase-pays-too-much error which we fixed
- windows installer puts binaries in a new location (daemon directory) and we correctly look for sysgeth there on startup
You should just install and run, if you were previous stuck on a block you can reindex to fix.
Syscoin v4.4.0 Release
Mandatory Upgrade
Syscoin v4.4.0 is now available.
Official block height for PODA/V19 fork is on block 1586000. Please upgrade before the height
Please see release notes for more detail:
https://github.com/syscoin/syscoin/blob/master/doc/release-notes/release-notes-4.4.0.md
For masternode operators, please upgrade your sentinel to v4.4.0 as well
Note chainlocks are disabled until the fork to avoid nodes banning each other between the two versions of nodes
Syscoin v4.4.0rc14 Testnet Only
if you are on macos with m1/m2 or ARM linux or aarch64 linux:
disable sysgeth in syscoin.conf by putting in command zmqpubnevm=
Note: This release candidate (rc14) does not have a codesigned version for OSX/WIN