tl;dr
Polygon's proof-of-stake network experienced a one-hour outage on July 30 that disrupted applications and users. While the CEO claimed block production continued, reports indicated a halt at block 74,592,238. The issue stemmed from a hotfix and temporary pause in the consensus layer, Heimdall, durin...
Polygon’s proof-of-stake (PoS) network experienced a one-hour outage on July 30 that disrupted applications and users. While Polygon CEO Sandeep Nailwal stated that block production continued uninterrupted, QuickNode Status reported a halt in block production at block height 74,592,238, a claim supported by multiple user observations on X.
Nailwal acknowledged that several RPC providers faced issues over a two-to-three-hour period, impacting corresponding applications and users. Despite these disruptions, the chain remained operational for unaffected RPCs, continuing to produce blocks and process transactions. Polygonscan data confirmed the network generated a new block two seconds after the stalled height, revealing a disparity between block production and explorer data.
The incident was traced to a hotfix and temporary pause in the consensus layer, Heimdall, related to a complex upgrade. The execution layer, Bor, operated normally, but some RPC nodes fell out of sync, causing application-level failures that mimicked a network halt. Nailwal apologized for the user impact and stated efforts were underway to synchronize providers and prevent further issues.
Infrastructure operators reported similar disruptions. QuickNode observed the mainnet stalling at block 74,592,238, linking the problem to the Heimdall v0.2.16 upgrade affecting Bor and Erigon execution clients. As a result, QuickNode halted upgrades and began node resynchronization and Erigon instance resets to restore service.
Polygon’s status page identified the Heimdall-V2 network as the fault origin. The mainnet’s Heimdall service became unresponsive, impairing validator and checkpoint visibility through Heimdall APIs, but without affecting the Bor layer. After diagnosing the issue, engineers deployed a fix and declared the incident resolved.
The timeline of events showed the incident was logged at 09:52 UTC, identified by 09:57, and resolved by 11:01. However, QuickNode reported a stall again at 11:28, paused the Heimdall upgrade rollout at 11:51, and resynced nodes by 15:39 to restore services. Nailwal described the event as a coordination failure between consensus and infrastructure teams rather than a protocol malfunction.