Follow-up to the previous stale-blacklist fix: with that fix live,
Android's balances_test.yml CI job progressed much further but still
hung for the full 25-minute job timeout. Live logcat + a full-log
retry analysis (grouping by host, comparing first/last timestamp
against the ~23 minute test window) showed 27 more chains whose nodes
were retried continuously for the ENTIRE run duration with zero
successful connections - as opposed to the ~90 other chains, which
connect once (or retry once or twice in the first ~2 minutes) and
then go quiet, i.e. genuinely healthy.
Every one of these 27 is a real, currently-dead third-party RPC
(Kusama Asset Hub, KILT, Crust Shadow, Phala, Mangata X, Joystream,
Mythos, Base, and 19 others) - not anything related to Tron or this
session's other changes. Each was matched to its current chain_id by
node URL against live chains.json, not guessed.
sync_from_nova.py's new stale-blacklist warning confirms all 33
entries (6 previous + 27 new) now match something in Nova's current
chain data - zero silently-ineffective entries this time.
This is inherently a moving target (third-party community RPC uptime
changes over time) - the warning added in the previous commit is what
makes that maintainable going forward instead of silently rotting for
another five months.
Root cause of the balances-test hang in pezkuwi-wallet-android (silent
for over an hour, cancelled multiple times before finding this):
every single chain_id in blocked-chains.json was stale - none matched
anything in Nova's current chains.json (last_updated 2026-02-11, ~5
months of upstream chain-id drift since). The blacklist mechanism
silently no-opped for all 6 entries, so none of the chains it was
meant to exclude were actually excluded.
With the blacklist ineffective, several chains with genuinely dead RPC
endpoints (confirmed live 2026-07-08: 3DPass, Curio, Quartz, Subsocial
- 502s, timeouts, SSL failures) stayed in the merged output. The
Android app's ChainConnection/NodeAutobalancer has no backoff/circuit
breaker for dead endpoints, so it retried them in an extremely tight
loop (~1 attempt/second, 1500-1800+ attempts observed in a 25 minute
window per chain) - burning resources indefinitely and starving the
app's actual chain-sync/test work, which is what made the Android CI
job hang for over an hour with zero progress.
Fix: re-verified and corrected all chain_ids (matching by current
name against live chains.json), added 3DPass/Curio/Subsocial which
weren't blacklisted before, dropped entries for chains no longer
present upstream at all (Passet Hub Testnet, Darwinia Crab, DeepBrain,
Exosama - nothing to block, kept them as historical noise otherwise).
Also added a stale-blacklist warning to sync_from_nova.py: it now
tracks which blocked chain_ids actually matched something in Nova's
current chains across all versions, and prints a warning listing any
that didn't - so this exact silent drift is caught at sync time going
forward instead of rotting unnoticed for another five months.
- pezkuwi-overlay/chains/pezkuwi-chains.json: add Tron mainnet
(native TRX + USDT-TRC20) chain entry, plus new TRX/Tron icons
under pezkuwi-overlay/icons/ (the correct source location - synced
into the top-level icons/ output by sync_from_nova.py).
- Regenerate all chains/*, xcm/*, staking/* merged output via
sync_from_nova.py against the current nova-base submodule state.
- Also picks up previously-uncommitted local fixes: canonical
Telegram/Twitter links (t.me/kurdishmedya, x.com/bizinikiwi) in
README/docs/banners, and Pezkuwi mainnet + People chain SubQuery
staking overrides.
- Created pezkuwi-overlay/config/global_config_overlay.json (source B)
with stakingApiOverrides for Pezkuwi SubQuery + prod API URLs
- Updated sync_from_nova.py with sync_config() function
Merges: nova-base/global + nova-base/staking (A) + overlay (B) → staking/ (C)
- Same merge pattern as chains, XCM, and icons
- Add staking, staking-rewards, history externalApi endpoints (subquery.pezkuwichain.io)
for both Pezkuwi Relay and Asset Hub
- Add identityChain reference to People chain on Relay and Asset Hub
- Add stakingWiki link to wiki.pezkuwichain.io/staking
- Add pushSupport option to Asset Hub
- Add defaultBlockTime to Asset Hub additional config
- Regenerate all merged chain versions via sync script
- Add blocked-chains.json with DNS-failing endpoints (AlephZero, InvArch, Quartz, Passet Hub Testnet)
- Update merge-chains.py to filter PAUSED chains, testnets (except Pezkuwi), and broken RPCs
- Reduce chain count from 102 to 86 (4 Pezkuwi + 82 working Nova chains)
- This fixes staking page not loading due to DNS resolution failures
Source changes (pezkuwi-overlay/chains/pezkuwi-chains.json):
- Pezkuwi Relay Chain: Add stakingMaxElectingVoters: 22500
- Pezkuwi Asset Hub HEZ: Change staking from ["nomination-pools"] to ["relaychain"]
- Pezkuwi Asset Hub: Add stakingMaxElectingVoters: 22500
This fixes the "Loading staking info" issue where findStakingTypeBackingNominationPools()
was failing because nomination-pools had no backing staking type.
Regenerated all chain versions via sync_from_nova.py
Asset Hub uses parent relay chain for era calculations since
it doesn't have Babe consensus module (uses Aura instead).
This fixes the crash when accessing nomination pools staking.
- Asset Hub HEZ staking set to null (was nomination-pools)
- Relay chain HEZ keeps relaychain staking only
- Wallet code prepared for Polkadot 2.0 but config disabled for now
pezkuwichain